CEO Connectedness and Corporate Frauds

Vikramaditya Khanna, E. Han Kim, and Yao Lu†

Abstract

This paper identifies an important factor in assessing the likelihood of corporate wrongdoing—the connection a CEO has with her top executives developed through their appointment decisions during her tenure. A sample of publicly-listed firms over the period 1996-2006 reveals that CEO connectedness within executive suites increases the likelihood of committing frauds and decreases the likelihood of detecting frauds. We identify several channels through which CEO connectedness decreases expected costs of committing fraud--by hindering and delaying detection, by reducing the likelihood of CEO dismissal upon fraud discovery, and by lowering the coordination costs of carrying out illegal activities. Furthermore, standard monitoring mechanisms do not seem to mitigate the adverse effects on frauds, suggesting regulators, investors, and governance specialists should pay particular attention to the CEO’s influence over top executives developed through appointments decisions. First Draft: August 22, 2011 This Draft: June 22, 2012 Keywords: Corporate Frauds, Social Connections, Corporate Governance, CEO Influence JEL Classifications: G30, K20 †Vikramaditya Khanna is Professor of Law at the University of Michigan Law School, Ann Arbor, Michigan 48109: vskhanna@umich.edu. E. Han Kim is Fred M. Taylor Professor of Finance at the University of Michigan Ross School of Business, Ann Arbor, Michigan 48109: ehkim@umich.edu. Yao Lu is Research Fellow at Mitsui Life Financial Research Center at the University of Michigan Ross School of Business: yaolu@umich.edu, and Assistant Professor of Finance at Tsinghua University School of Economics and Management, Beijing, China: luyao@sem.tsinghua.edu.cn. We have benefited from helpful comments and suggestions by seminar participants at the University of Michigan Law School and Ross School of Business, University of Southern California Law School, University of Miami Law School, Central University of Finance and Economics, Renmin Univesity, Indian School of Business, Hyderabad, and Shanghai University of Finance and Economics. We thank Joseph Rewoldt, Satoko Kikuta, Guodong Chen, Paul A. Sandy, Alex Pierson, Melan A. Patel, and Robert Powell for excellent research assistance. This project received generous financial support from Mitsui Life Financial Research Center at the University of Michigan.

2

Accusations of corporate and securities fraud have dominated headlines over the last decade.

Corporate wrongdoing damages investor confidence, hurts shareholder value, causes misallocation of

capital, and increases financial market instability,1 leading a number of financial economists to examine

factors affecting the likelihood of fraud and its detection. The list of factors includes board structure

(Beasley, 1996; Agrawal and Chadha, 2005), general business conditions (Povel, Singh, and Winton,

2007; Wang, Winton, and Yu, 2010), corporate lobbying (Yu and Yu, 2011), market- and regulatory-

based institutions (Dyck, Morse, and Zingales, 2010), and executive compensation (Hertzberg, 2005;

Burns and Kedia, 2006; Efendi, Srivastava, and Swanson, 2007; Peng and Röell, 2008).

Noticeably absent in these inquiries is the role CEOs’ personal connections with, and influence

over their top executives play in corporate frauds. CEOs have both explicit legal authority and substantial

“soft” influence within the firm to direct corporate behavior,2 of which wrongdoing is but one potential

outcome. In this paper we focus on the “softer” side of internal governance practiced in executive

suites—CEO influence over other top executives built through appointment decisions. We ask: How does

CEO connectedness within executive suites affect the likelihood of wrongdoing and its detection?

The answer to this question is not obvious. As in other corporate activities, corporate wrongdoing

often requires coordination between, or acquiescence by, top executives who report to CEOs on a daily

basis.3 The coordination/acquiescence may be in the form of direct involvement in criminal activities or a

1 See Karpoff and Lott (1993); Beatty, Bunsis, and Hand (1998); Bhagat, Bizjak, and Coles (1998); Karpoff, Lee, and Vendrzyk (1999); Bar-Gill and Bebchuk (2002); Karpoff, Lee, and Martin (2008a, 2008b, 2010); Gande and Lewis (2009); and Murphy, Shrieves, and Tibbs (2009). 2 Allen, Kraakman, and Subramanian (2011) discuss CEOs’ legal authority to contractually bind the firm for ordinary transactions. Evidence on the importance of CEO influence on firm behavior and performance is provided in Bertrand and Shoar (2003) who find CEO fixed effects matter for a wide range of firm policies; Bennedsen, Perez-Gonzalez, and Wolfenzon (2006) who document that CEO deaths are strongly negatively correlated with firm profitability and growth; Cronqvist, Makhija, and Yonker (2012) who show differences in corporate financial leverage can be traced to CEOs’ personal leverage; and Jenter and Lewellen (2011) who find CEO age approaching retirement has an important impact on the likelihood of their firms being taken over and the takeover premiums their shareholders receive. 3 We focus on top executives rather than outside board members. In theory, one could argue that outside board members do have a role in corporate wrongdoing because their key role is to monitor management. However, outside board members, who work on a part-time basis, are rarely involved in the day-to-day activities of the firm and, hence, are rarely held liable for failing to monitor behavior at the firm when they are not themselves actively involved in, and benefit from, the wrongdoing (Black, Cheffins, and Klausner, 2006a), Steven Davidoff, “Despite Worries, Serving at the Top Carries Little Risk,” The NEW YORK TIMES, June 7, 2011). Of course, there are the rare and notable exceptions, such as the outside board members of Enron and Worldcom who paid substantial penalties (Ben White, “Former Directors Agree To Settle Class Actions,” WASHINGTON POST,

3

reluctance to blow the whistle if an executive becomes aware of fraud. Thus, a CEO’s close

connectedness with his top lieutenants may facilitate wrongdoing. However, the relationship may also

help deter corporate frauds. The CEO’s familiarity with other top executives may enable him to detect

early signs of fraud. Or when a CEO is unaware or uncertain about the illegality of certain activity, a

common problem in some areas of white collar crime, closer interpersonal relationships could make it

easier for other executives to provide friendly information to the CEO to help avoid wrongdoing.

To examine whether and how CEO connectedness within executive suites affects corporate frauds

involving management, as a proxy for CEO connectedness we use relationships a CEO develops with top

executives through appointment decisions during her tenure. We find this measure of connectedness

increases the likelihood of fraud involving management by decreasing the expected costs of committing

frauds. Specifically, CEO connectedness decreases the likelihood of fraud detection, lengthens the time

from fraud incidence to its detection, reduces the likelihood of forced CEO turnover upon discovery of

fraud, and lowers the coordination costs needed to carry out illegal activities. In addition, we find these

effects are not mitigated by standard monitoring and governance mechanisms.

We measure CEO connectedness within executive suites by the fraction of top executives appointed

(FTA) to the list of top four non-CEO executives during a CEO’s tenure. (The list is from ExecuComp,

which ranks executives by the sum of salaries and bonuses.). CEOs are heavily involved in identifying,

recruiting, and appointing top executives and in deciding their compensation. If an executive making the list

is a new hire or newly promoted within the firm, she is more likely to share similar beliefs and visions with,

and may be beholden to, the CEO who hired or promoted her than executives appointed by a previous CEO

(Landier, Sraer, and Thesmar, 2008). An executive getting a big enough raise in salary and bonuses to leap

onto the list may be similarly aligned with, and beholden to the CEO. Thus, a higher FTA increases a

CEO’s internal influence in executive suites through what social psychologists refer to as social influence

January 8, 2005). The rather exceptional circumstances of these settlements are discussed in Black, Cheffins, and Klausner (2006b) and indicate that such settlements are a rarity, not the norm.

4

(Cialdini, 1984), which rely on norms of reciprocity, liking, and social consensus to shape management’s

decision making. The number of top executives appointed during a CEO’s tenure at a particular point in

time, however, depends on factors such as the length of the CEO’s and other executives’ tenure (Landier et

al., 2008). Thus, for robustness we use the abnormal fraction of top executives appointed (AFTA), the

residual of a regression relating FTA to factors correlated to FTA. All tables throughout the paper report

estimation results based on both FTA and AFTA.

Our full sample covers 17,797 firm-year observations associated with 2,736 unique firms during

the period 1996 through 2006. For this sample, we gather comprehensive data on alleged corporate frauds

and apply carefully designed screens to exclude mistaken or frivolous suits, identifying 282 cases of

corporate fraud with 684 firm-year fraud observations.

Inherent in any sample construction of frauds is the partial observability problem: We observe

only detected frauds, not the population of frauds. Since observed fraud depends on two distinct but latent

processes—commitment of fraud and detection of fraud—we follow Wang et al. (2010) and Wang (2011)

and employ the bivariate probit model. We find both FTA and AFTA are positively related to the

likelihood of wrongdoing and negatively related to the likelihood of detection given wrongdoing. Our

estimates indicate that a firm with all top four executives appointed during the CEO tenure (FTA = 1) has

34.4% higher fraud incidence and 20.3% lower likelihood of detection given fraud than a firm with none

of its top four executives appointed during the CEO’s tenure (FTA = 0).

We address endogeneity issues by estimating instrumental variable (IV) regressions, using as our

IVs the number of non-CEO top executive deaths and permanent retirements at the age of 70 or older.

Deaths and retirements of top executives appointed by previous CEOs automatically increase FTA and

AFTA, but they are unlikely to be related to fraud given our selection criteria. The IV regression

estimations demonstrate the robustness of the CEO connectedness-fraud relation. It is also robust to major

organizational changes that may cause top executive turnovers, clustering standard errors at different

levels, alternate measures of CEO connectedness, and alternate sample constructions.

5

We also investigate channels through which CEO connectedness affects wrongdoing by focusing

on factors directly related to the expected costs of fraud: the probability of detection, the time taken to

detect fraud, the likelihood of forced CEO turnover upon discovery of wrongdoing, and the coordination

costs of conducting wrongdoing.

Closer CEO connectedness may help conceal fraud. This might happen via influencing others to

fabricate or obfuscate internal records, thereby making it harder to detect or prove wrongdoing in court

(Arlen and Carney, 1992; Khanna, 2003), or by simply pressuring individuals to conceal instances of

wrongdoing out of loyalty to the CEO who appointed them, or out of fear of reprisal. Bivariate model

estimations reveal FTA and AFTA are negatively related to detection, given fraud. FTA and AFTA also

are positively related to fraud detection duration, the period from the commencement of fraudulent

activity to the filing of the class action litigation, and negatively related to the Cox-hazard ratio of fraud

detection. The estimated coefficient of FTA implies a fraud by a firm with all top four executives

appointed during the CEO’s tenure (FTA = 1) will take 348 days longer to be detected than a fraud by a

firm with none of its top four executives appointed during the CEO’s tenure (FTA = 0).

CEO connectedness may also help avoid CEO dismissal upon discovery of wrongdoing. Detected

frauds do not automatically lead to a forced CEO turnover. Unless a CEO is sentenced to a jail term,

dismissal could be largely a firm level decision, which may be affected by CEO connectedness. More

connected CEOs may be able to garner greater support from executive suites (to persuade the board of

directors) to retain their positions than less connected CEOs. Consistent with this prediction, we find

closer CEO connectedness is associated with a lower forced CEO turnover-fraud sensitivity. Our

estimates indicate that the probability of forced CEO turnover following a fraud by a firm with FTA = 1 is

29.6% lower than that by a firm with FTA = 0.

In addition, when CEOs and top executives are more closely connected, it is easier to obtain

support to override internal control mechanisms or push through policies or activities that others on the

executive team may be reluctant to pursue (Khanna, 2003); in other words, coordinating illegal activities

6

is less costly. With lower coordination costs, more executives are likely to get involved in frauds and

charged when detected. That is what we find; the number of executives charged with fraud is positively

and significant related to CEO connectedness.

We estimate the two-stage IV regressions for each of the three channels through which CEO

connectedness reduces the expected costs of frauds. All relations are robust.

Finally, we ask whether the adverse effects of CEO connectedness can be contained by standard

monitoring mechanisms. We re-estimate the bivariate probit model while interacting FTA or AFTA with

proxies for the strength of internal monitoring by the board and audit committee. We also interact FTA or

AFTA with the strength of external monitoring as proxied by institutional ownership concentration. None

of the interactive effects is significant. The standard monitoring mechanisms do not seem effective in

countering the influence of CEO connectedness on wrongdoing.

This paper contributes to the literature on determinants of corporate fraud involving management.

We extend previous studies by identifying a new determinant of both the incidence of fraud and its

detection—CEO connectedness within executive suites.

Our paper is also related to the literature on social and network connections, which explores the

effects of the connectedness on the selection of the higher levels of the corporate hierarchy (e.g., CEOs,

board members), various indicia of performance, interlocking board directorships, and fund performance.4

We focus on a different type of connection, the connectedness CEOs develop with their top executives

through their appointment and compensation decisions, and illustrate how this type of connection affects

the incidence of corporate wrongdoing and its detection.

Since CEO connectedness within executive suites enhances CEOs’ internal influence, this paper

also adds new evidence on the impact of CEO power. Adams, Almeida, and Ferreira (2005) find firms

with powerful CEOs have greater variation in performance; Landier et al. (2008) and Bebchuk, Cremers,

4 See, for example, Shivdasani and Yermack (1999); Adams and Ferreira (2007); Hochberg, Ljungqvist, and Lu (2007, 2010); Hwang and Kim (2009); and Fracassi and Tate (2011); Ferris, Jagannathan, and Pritchard (2003); Larcker, Richardson, Seary, and Tuna (2005); Stuart and Yim (2010); and Cohen, Frazzini, and Malloy (2008, 2010).

7

and Peyer (2011) find CEO power is associated with lower firm performance; Bebchuk and Fried (2004)

find powerful CEOs reduce the link between compensation and performance; Morse, Nanda, and Seru

(2011) show powerful CEOs often rig incentive contracts. Our paper finds CEO power—stemming from

CEO connectedness—is also conducive to corporate wrongdoing.

The rest of the paper is organized as follows. The next section contains the empirical design, data

description, and summary statistics. Section II provides the main results. Section III explores the channels

through which CEO connectedness influences the likelihood of wrongdoing. Section IV investigates

whether standard monitoring mechanisms help contain the adverse effects of CEO connectedness. Section

V concludes.

I. EMPIRICAL DESIGN, DATA, AND SUMMARY STATISTICS

I.1 Empirical Methodology

In estimating the relation between CEO connectedness and fraud, we are mindful of the partial

observability problem: We observe frauds only when they are detected. To address this issue, we follow

Wang et al. (2010) and Wang (2011) and employ the bivariate probit model. For each firm i, we denote

Fraudit* and Detectit

* as the latent variables determining firm i’s likelihood of committing a fraud in year t

and the possibility of detecting it as follows:

Fraudit* = XF,it it (1a)

Detectit* = XD,it it (1b)

XF,it is a vector of variables explaining firm i’s likelihood of committing a fraud in year t, and XD,it

contains variables explaining the firm’s likelihood of being detected. it and it are zero-mean disturbances

with a bivariate normal distribution. The correlation between it and it is .

We define the following binary variables: Fraudit = 1, if Fraudit* > 0, and Fraudit = 0, otherwise;

and Detectit = 1 if Detectit* > 0, and Detectit = 0, otherwise. We do not directly observe the realizations of

Fraudit and Detectit; instead, we observe Observeit = FrauditDetectit, where Observeit = 1 if firm i has

8

committed fraud and has been detected, and Observeit = 0 if firm i has not committed a fraud or has

committed fraud but has not been detected.

Let denote the bivariate standard normal cumulative distribution function. The empirical model

for Observeit is:

P(Observeit = 1) = P(FrauditDetectit = 1) = XF,it , XD,it , ) (2a)

P(Observeit = 0) = P(FrauditDetectit = 0) = 1 - (XF,it , XD,it , ) (2b)

Thus, the log-likelihood function for the model is:

L( log(P(Observeit = log(P(Observeit = 0)) (3)

This model can be estimated using the maximum-likelihood method.

An important assumption of the bivariate probit model is that XF,it and XD,it do not contain the

same set of variables such that at least one vector has variables not present in the other vector. As we

discuss below, this condition is satisfied in our study because some variables affect fraud incidence

directly without appreciably affecting the likelihood of detection, yielding variables in XF,it that are not

present in XD,it. Further, because the bivariate probit model estimation assumes no false detections, we

cull likely false detections by exercising great care in assembling our sample of detected frauds, as

described in the next section.

To account for possible correlations among firms in the same industry, robust standard errors are

clustered at the industry level, which are defined as Fama-French (1997) 48 industry groupings. The

results are robust to clustering standard errors at the firm and the CEO-firm level, which are reported in

the Appendix (Table A-7).

I.2 Variables

I.2.1 Identifying Fraud

We construct a sample of frauds involving management by using two data sources: private

securities fraud class action lawsuits and the SEC litigation releases. The private securities fraud class

action data source is the Stanford Securities Class Action Clearinghouse (SSCAC), which provides a

9

thorough collection of likely discovered fraud cases, including virtually all frauds with more than a de

minimis effect on stock price that could generate private litigation.5 Filing a securities class action lawsuit

is now a largely automated process whereby law firms file a suit whenever there is a negative stock price

movement above a certain de minimis level (Choi, Nelson, and Pritchard, 2009).

We begin the sample period from 1996, the year after the passage of the Private Securities

Litigation Reform Act, which was designed to reduce frivolous suits by requiring greater and more

credible evidence to be provided earlier in the litigation process before a suit would be permitted to

proceed to the next stage of litigation, e.g., “discovery” (Johnson, Nelson, and Pritchard, 2007). However,

using the SSCAC database still raises the specter of false detections, which motivates Dyck et al. (2010)

to apply careful screening procedures to exclude suits that could potentially be mistaken or frivolous. We

follow their screening procedures with some modifications. Specifically, we exclude: (1) cases where the

SSCAC identifies regular agents other than management as initiating the fraud—to identify frauds

involving top management; (2) cases subsequently dismissed by the court; and (3) settled cases where the

settlement amounts are less than $3 million. The threshold of $3 million originates from previous studies

(Grundfest, 1995; Choi, 2007; and Choi et al., 2009), which suggest a settlement amount as an indicator

to separate frivolous suits from meritorious ones. They find suits settling below a $2.5 - $1.5 million

threshold are on average frivolous.6 These screens are the standard treatment in the securities fraud

literature when addressing the concern of over-inclusion. They are generally seen to whittle away the

most likely types of frivolous or mistaken suits, leaving us primarily with cases of likely fraud.7 Unlike

Dyck et al., however, we do not exclude backdating cases, IPO underwriter allocation cases, mutual fund

timing and late trading cases, analyst cases involving false provision of favorable coverage, or cases

5 The SSCAC maintains and updates records of federal securities fraud class actions since the 1995 passage of the Private Securities Litigation Reform Act. The database is available at http://securities.stanford.edu/index.html. 6 The range reflects the cost to the law firm for its effort in filing. A firm settling for less than $1.5 million is almost certainly just paying lawyer fees to avoid negative court exposure. 7 Even if all cases led to judicial determinations with no settlements, one still would not know for certain the total number of actual fraud cases. A judicial determination of liability could also be erroneous given that it often requires assessing whether someone had a particular state of mind – an inherently difficult matter on which to have certainty. The potential presence of a few non-fraud suits in the fraud sample adds noise, weakening the power of our test to find significant results.

10

where defendants claim they are settling to avoid negative publicity if the settlement amount exceeds $3

million.

Information about the SEC’s enforcement actions is hand collected from the SEC litigation

database (http://www.sec.gov/litigation.shtml), which contains information on civil lawsuits and

administrative proceedings brought by the SEC for alleged financial misreporting, insider trading, and

violations of the Foreign Corruption Practice Act and Sarbanes-Oxley Act, as well as other alleged

violations of the Federal Securities Laws and accompanying regulations. The SEC enforcement actions

are thought less likely to be frivolous or mistaken than private suits because the people making

enforcement decisions (the SEC employees) do not directly receive the monetary remedies from such

suits (as a private litigant might). The SEC may pursue a case involving small damage, resulting in a

small settlement, because the case raised important legal or enforcement questions, or demand substantial

changes in financial reporting and corporate governance in lieu of a large settlement. Therefore, we do not

apply the $3 million screen on the settlement amount for SEC enforcement actions. However, we retain

the requirements that the cases involve top management and are not dismissed by the court.

While the SSCAC provides precise beginning and ending dates of fraudulent activities, and filing

dates of class actions, which we use as detection dates, the SEC litigation releases are often ambiguous

about when a fraud started or was detected. The SEC database only occasionally mentions the quarter or

month when an alleged fraud began and is ambiguous about detection dates. Consequently, estimation of

how long it took to detect a fraud relies only on the private litigation cases in the SSCAC.

For both private class action and the SEC cases, we include cases filed between 1996 and 2009

making allegations of fraud occurring between 1996 and 2006. The sample period stops at the end of

2006 to provide a three-year detection period through 2009. (Our data compilation started in 2010.) The

three-year period allows inclusion of frauds that took place during the sample period but were detected

and settled between 2007 and 2009. In our sample of private suits, the average duration from the

commencement of fraudulent activity to the filing of a class action is 760 days. For firms that had

11

multiple securities lawsuits, we use only the one involving the most number of executives charged in the

litigations. Our screening criteria yield 282 fraud cases with 684 fraud firm-year observations.

I.2.2 Measures of CEO Connectedness

We estimate CEO connectedness by calculating the fraction of top four non-CEO executives

appointed, FTAit, the number of appointees during the tenure of firm i’s CEO as of year t divided by four.

ExecuComp ranks executives by the sum of salaries and bonuses. To prevent changes in the reported

number of executives from adding noise to within-firm variation in FTAit, we drop firm-year observations

when ExecuComp reports less than four top non-CEO executives. 8 We assume the year a non-CEO

executive first appears on the list of top four non-CEO executives is the year in which she obtained the

position. We compare this year with the year the current CEO took office to determine whether the

executive is appointed during the current CEO’s tenure.

Landier et al. (2008) show the fraction of top executives hired by the current CEO’s tenure is

correlated with the length of the CEO’s tenure, the average tenure of non-CEO top executives, and whether

the CEO is hired from outside the firm. Thus, for robustness, we use an alternative proxy for CEO

connectedness, the abnormal fraction of top executives appointed (AFTA) during a CEO’s tenure.9 AFTAit

is the residual of regression (4):

FTAit= a0+ a1CEO_Tenureit+ a2Execsenit + a3Outsideit+ a4Unknownit+ a5FTA_1Yit

+ a6FTA_1Y_Unknownit + Yeart + it (4)

CEO_Tenureit is the number of years firm i‘s CEO has been in office by year t.10 Execsenit is the average

number of years firm i’s top four non-CEO executives have held their positions by year t. Outsideit is an

8 Kim and Lu (2011) illustrate the importance of keeping the number of executives constant when constructing executive variables for panel regressions with firm fixed effects. Cross-checking against proxy statements shows that missing executives in ExecuComp are due to omission: The firm-year observations with less than five top executives in ExecuComp show five or more top executive in proxy statements. 9 Landier et al. (2008) use a similar measure based on new hires only. Since similar connections may arise between a CEO and top executives through promotion within the firm and/or compensation increases, Kim and Lu (2012) include all top executives added to the list of top four non-CEO executives during the CEO’s tenure. 10 If a CEO leaves the position and returns later, ExecuComp reports only the latest appointment date. Thus, simply comparing the CEO appointment date reported by ExecuComp with the current year may generate negative CEO tenure. We correct for this problem by backtracking the previous appointment year using the CEO and company names.

12

indicator variable equal to one if a CEO is from outside the firm. Unknownit is the fraction of executives

whose first year on the list of the top four non-CEO executives cannot be identified based on data provided

by ExecuComp. This variable is designed to control for noise in FTAit and Execsenit due to ambiguity about

the precise year in which some of the top executives were appointed.11 FTA_1Yit is the fraction of top

executives appointed during a CEO's first year in office (a new CEO appointment is sometimes followed by

several top executive turnovers). To control for noise in FTA_1Yit, we add FTA_1Y_Unknownit, the fraction

of top executives whose appointment cannot be determined as occurring during a CEO's first year in office.

The regression also controls for year-fixed effects to account for macroeconomic factors affecting

top-executive hiring, promotion, and retention decisions. The regression estimate is reported in Table A1

in the Appendix. Unsurprisingly, the fraction of executives appointed during a CEO’s tenure is positively

related to the length of the CEO tenure and negatively related to the average non-CEO executive tenure.

I.2.3 Control Variables

Estimating the bivariate probit model requires two sets of control variables, one each for (1a) and

(1b). They may overlap, but should not be identical. Thus, some of the variables explaining fraud

incidence should be different from those explaining fraud detection. Since the expected cost of

committing fraud depends on the probability of detection, all variables affecting fraud detection should

also affect fraud incidence. However, the reverse may not be true; factors affecting fraud incidence may

not have obvious implications for the likelihood of detection, especially if the parties responsible for

detection do not rely on the fraud incidence factors when attempting to detect wrongdoing. Thus, all

control variables for fraud detection are also control variables for fraud incidence, but fraud incidence has

additional control variables. We first describe the set of control variables for fraud detection and follow

with a second group of control variables that apply only to fraud incidence.

11 ExecuComp provides appointment dates for CEOs, but not for other top executives (except for the CFO beginning in 2006). Hence, if an executive is already one of the top four non-CEO executives when the firm first appears in ExecuComp, we cannot determine when he first obtained the position. For such cases, we use the year the executive first appears in ExecuComp as the year of appointment to the top executive position and compare it with the year the current CEO took office to determine whether the executive was appointed during the CEO’s tenure. Since this method underestimates FTA, we include Unknown as a control variable to mitigate the underestimation problem.

13

Fraud Detection

Internal and external monitoring may play an important role in detecting fraud. Control variables

related to internal monitoring include: (1) The percentage of non-independent directors on the

board, %_NonIndepDirectors. The monitoring role played by independent directors has been widely

documented; for example, Weisbach (1988) finds CEO turnover following poor performance is positively

related to the fraction of outside directors. (2) Log of the number of directors on the board, Ln(BoardSize).

Prior research indicates larger boards tend to be less effective monitors (Lipton and Lorsch, 1992; Jensen,

1993; Yermack, 1996; and Eisenberg, Sundren, and Wells, 1998). (3) Log of the number of board

meetings in a given year, Ln(BoardMeetings), which may indicate the strength of board oversight and

monitoring (Vafeas, 1999). (4) The percentage of non-independent directors on the audit

committee, %_NonIndepDirectors_Audit, and (5) the log of the number of directors on the audit

committee, Ln(AuditComSize). Audit committees, charged with the oversight of financial reporting,

internal controls, and external audits, play an important role in fraud detection (Deli and Gillan, 2000).

The strength of external monitoring is proxied by institutional ownership concentration (IOC) and

analyst coverage. Previous studies document the important roles institutional investors play in shaping

corporate governance (e.g., Hartzell and Starks, 2003; Cremers and Nair, 2005; Del Guercio, Seery, and

Woidtke, 2008; Edmans, 2009; Kim and Lu, 2011). We follow Hartzell and Starks (2003) and estimate

IOC by the percentage shareholdings of the top five institutional investors. Analyst coverage is also

considered as an important form of external monitoring, reducing information asymmetry (e.g., Hong,

Lim, and Stein, 2000; Brav and Lehavy, 2003; Chang, Dasgupta, and Hilary, 2006; Das, Guo, and Zhang,

2006; and Kelly and Ljungqvist, 2011). Analyst coverage, Ln(Analyst), is measured as the logged value of

one plus the number of analysts following a firm in a given year.

Because our sample of detected frauds is obtained from lawsuits, we control for firm- or industry-

specific litigation factors. The securities litigation literature (e.g., Jones and Weingram, 1996; Johnson et

al., 2007) suggests that firm performance and stock return volatility are related to a firm’s litigation risk.

14

Firm performance is proxied by Tobin’s Q and Ebitda/TA. Tobin’s Q is measured as the ratio of the

market value of common equity plus the book value of total liabilities to the book value of total assets.

Ebitda/TA is measured as earnings before interest, taxes, depreciation, and amortization divided by the

book value of total assets. Stock return volatility, StockVolatilities, is measured as the standard deviation

of daily stock returns over a given year.

Litigation intensity can be correlated among firms within the same industry, and high industry

litigation intensity may affect an individual firm’s litigation risk (Wang et al., 2010). We control for

industry securities litigation intensity, IndustryLitigation, the number of all unscreened lawsuits against

publicly-listed firms in an industry in a given year divided by the number of firms covered by Compustat

in the same industry in the same year.

We also include firm size, sales growth rate, and leverage. Larger firms and higher growth firms

tend to attract more investor attention, leading to greater public scrutiny. Wang et al. (2010) show frauds

by larger firms are more likely to be detected than frauds by smaller firms. Banks and fixed-income

investors may monitor firms more closely when firms have high financial leverage. Firm size is measured

as log of the book value of total assets, Ln(TotalAssets); growth rate is measured by SalesGrowth_3Yr,

the three-year annual growth rate in sales as reported in ExecuComp; financial leverage, Leverage, is the

sum of short- and long-term debt divided by the book value of total assets.

Finally, we control for CEO age, Ln(CEO_Age); whether a CEO chairs the board, CEO_Chair;

an indicator for a founder-CEO, CEO_Founder; and CEO tenure, CEO_Tenure. We follow Bebchuk et al.

(2011) and classify a CEO as a founder if he was the CEO five years prior to going public, where the date

of going public is assumed to be the first date the firm appears in the CRSP database. Older CEOs tend to

be more experienced, which may help them evade detection. Chairing the board or being a founder gives

a CEO more power, making it easier to weaken the intensity of internal and external monitoring. The

CEO/chair position and CEO tenure may also reflect the CEO’s ability and performance.

Fraud Incidence

15

To the extent these variables affect the likelihood of detection, they will also affect the likelihood

of committing frauds by affecting the expected cost of committing fraud. Some of these variables may

also directly influence fraud incidence by affecting the incentives to commit fraud. For example, Tobin’s

Q and Ebitda/TA may be negatively related to fraud incidence, as fraud is more likely when a firm is

suffering operating troubles (Arlen and Carney 1992; Alexander and Cohen, 1999); greater litigation

intensity may discourage frauds; higher leverage may provide the incentive to inflate earnings and other

accounting numbers to avoid violating debt covenants; career concerns may discourage younger CEOs

from committing frauds; and founder CEOs or CEOs with longer tenure may have more opportunities or

time to commit fraud than non-founders or CEOs with shorter tenure.

Some variables affect the likelihood of committing frauds without an obvious influence on

detection. Wang et al. (2010) argue the incidence of fraud is related to investor beliefs about industry

prospects and provide evidence of a hump shaped relation with industry Tobin’s Q. We control for

Industry Q and (Industry Q)2. CEO share ownership may also affect the incidence of frauds by directly

influencing CEO incentives and control rights, in turn, impacting firm performance and risk taking. Kim

and Lu (2011) show CEO share ownership, CEO_OWN, is related to firm performance and risk taking in

a hump shaped fashion. We control for CEO_OWN and (CEO_OWN)2. Finally, product market

competition affects the strength of governance (Guadalupe and Wulf, 2010; Giroud and Mueller, 2010;

Kim and Lu, 2011), which may in turn affect fraud incidence. We control for industry concentration ratio

(ICR) as a proxy for the competitiveness of product markets. ICR is the sum of the market share of the

four biggest firms in sales among all firms in Compustat in the same industry in a given year.12 A lower

ratio indicates greater competition. There is little reason to believe these factors affect the likelihood of

detection. For example, likely enforcers such as the SEC do not seem to rely on these factors in deciding

12 The Economic Census uses the largest 4, 8, 20, or 50 companies to compute ICRs. Because Compustat covers only public firms, we rely on the four largest companies to minimize the possibility of excluding private firms.

16

their detection efforts (Cox, Kiku and Thomas, 2003; Bowen, Call, and Rajgopal, 2010); instead, the SEC

seems to focus on other factors such as financial distress (Cox et al., 2003).

I.3. Sample Construction

Executive data are taken from ExecuComp; firm characteristics and accounting data, from

Compustat; stock return data, from CRSP; board information, from RiskMetrics; analyst coverage data,

from I/B/E/S from Thomson Reuters; and institutional ownership data, from the CDA Spectrum database.

Fraud data is manually constructed from the class action securities fraud litigation database by the

SSCAC and hand collected from the SEC’s litigation releases in its website. Merging these databases

provides a large panel dataset from 1996 through 2006.

Table II reports the sample distribution of firms that have data available to construct FTA

(Column 1) and of firms with identified fraud (Column 2). Panel A shows the distribution by year. The

number of sample firms is relatively stable over time, while the number of firms with frauds varies. In the

first few observation years, fraud firms are small both in number and percentage, but increase as the year

progresses. The peak is reached in 2001 and 2002, the two years with an unusually large number of

business scandals. The year of fraud is defined as the year when fraud took place, not the year of

detection. When a fraud lasts more than one year, we have multiple firm-year observations associated

with that fraud. Panel B shows the sample distribution by FTA. The percentage of firms with fraud

increases with FTA, suggesting a positive correlation between FTA and fraud.

I.4 Summary Statistics

Table III contains summary statistics for all key variables. The statistics for the full sample are

reported in Panel A. The mean Fraud is 0.038, indicating fraud observations account for 3.8% of all firm-

year observations. On average, it takes about 760 days from the commencement of fraudulent activity at

the firm to the class action filing. The majority of frauds (72%) involve accounting related matters, while

27% and 25% of frauds involve cases in real business activities and executives taking advantages of their

positions. (Some frauds belong to multiple types.) The mean and median FTA is 0.41 and 0.5, suggesting

17

that in a typical firm-year, about half of the top four non-CEO executives are appointees of the current

CEO. The mean and median of AFTA, the regression residual, are close to zero.

Panel B reports the mean of each variable separately for the fraud and non-fraud sample in

Columns (6) and (7), followed by Columns (8) and (9) showing the difference in the means and the P-

value of the t-test for the difference. The fraud sample shows significantly higher values for all four

measures of CEO connectedness than the non-fraud sample. Many control variables are significantly

different between the fraud and non-fraud sample. However, it is difficult to draw inferences on how

these variables are related to the likelihood of fraud from these univariate comparisons because many of

the variables are related to both fraud incidence and detection and frauds are observed only when detected.

Table IV presents pair-wise correlations between the variables of main interest. Panel A is based

on all firm-year observations. All four measures of CEO connectedness are highly correlated with each

other. They also are all positively and significantly correlated with the fraud indicator. Panel B is based

on cross sectional data on fraud. Here FTA and AFTA are firm-year averages over the fraud duration

period. FTA is positively and significantly related to the number of days from commencement to detection

of fraud and the number of executives charged in litigation. AFTA’s correlations with these variables are

also positive but only its correlation with the number of executives charged is significant.

II. CEO CONNECTEDNESS AND CORPORATE FRAUDS

Table V reports the bivariate model estimation results. Columns (1) and (3) show the estimation

results for the likelihood of committing fraud; Column (2) and (4), the probability of detection, given

fraud. The coefficients on the variables of main interest, FTA and AFTA, show the predicted signs and are

significant. CEO connectedness is associated both with greater fraud incidence and lower detection. The

estimated coefficient of FTA in the fraud regression suggests that a firm with all top four executives

appointed during the CEO tenure (FTA = 1) has 34.4% higher fraud incidence and 20.3% less likelihood

of detection than a firm with no top four executives appointed during the CEO tenure (FTA = 0).

18

Because CEO connectedness within executive suites is endogenous, its relations with fraud

incidence or with the detection may not be causal. To infer causality, we construct instrumental variables

(IVs) related to FTA and AFTA but unrelated to corporate wrong doing. For this purpose, we use top non-

CEO executive turnovers due to death and permanent retirement unrelated to fraud during the current

CEO’s tenure, Num_Death and Num_Retire, respectively. Fracassi and Tate (2011) use similar death and

retirement variables to construct IVs for social ties between independent directors and CEO.

Table III shows, on average, 59% of the non-CEO top executives are previous CEOs’ appointees,

whose turnovers, by definition, lead to an increase in FTA. Hence, executive turnovers due to death and

retirement should be positively related to FTA and AFTA. To ensure the executives’ deaths are unrelated

to fraud, we search media articles from Factiva on the cause of the deaths at fraud firms. We find none

can be attributed to suicide. Retirement may be less exogenous; an executive who committed fraud or is

afraid of being tainted by scandals may resign to disassociate himself from the firm. Thus, we impose

very stringent conditions to prevent fraud-related retirements from entering our IV constructions.

Specifically, we require that (1) an executive is 70 years or older at the time of retirement and (2) the

retirement is permanent by excluding those whose names later reappear as executives in any firms

covered by ExecuComp. For a number of executives, ExecuComp shows a time lag between the year their

names last appear as an executive and the year of their departure from the company—i.e., a number of

executives are dropped from the list of top executives before they actually leave the firm. If the time lag is

two years or longer, we assume the executive was demoted prior to the departure and do not attribute the

turnover to death or non-fraud related retirement. Table III indicates that there are more deaths than

voluntary retirements during a CEO’s tenure, perhaps because of the stringent conditions used to define

voluntary retirement and there are not many people past the age of 69 still occupying a non-CEO top

executive position.

The first-stage estimation results of the IV regressions are reported in Table A2 in the Appendix.

The dependent variable is FTA in Columns (1) and (2); AFTA in Columns (3) and (4). Columns (1) and (3)

19

correspond to the fraud incidence regressions and, hence, include the IV variables and all control

variables used in the fraud incidence regressions in Table V. Columns (2) and (4) correspond to the fraud

detection regressions, including the IV variables and all control variables used in detection regressions.

When we estimate the bivariate model in the second stage, the error terms of the fraud incidence

regression and detection regression are allowed to be correlated. Thus, Columns (1) and (2) are estimated

together as a system of simultaneous equations, as are Columns (3) and (4). As expected, both death and

retirement related turnovers are positively related to both FTA and AFTA with death being highly

significant. F-statistics (IVs), representing an F-test that all the instrumental variables are jointly zero,

have coefficients that are all greater than 10.

The second-stage estimation results are presented in Table VI. The results are robust. FTA_Hat

and AFTA_Hat, the predicted values of CEO connectedness, are both positively and significantly related

to the incidence of fraud and negatively and significantly related to fraud detection.

Table VI also shows more control variables with statistical significance than Table V. The control

variables may be correlated with CEO connectedness, causing estimation bias. When we correct for

endogeneity concerning CEO connectedness through the two-stage IV regression, the control variables’

coefficients are also corrected.

As expected, frauds are less likely and given fraud, detection is more likely when firms are valued

higher, more profitable, more highly levered, larger, subject to more frequent board meetings, and have

greater institutional ownership concentration and stock return volatility. Also as expected, firms with

founder CEOs and a higher fraction of non-independent directors are more inclined to commit frauds and

less likely to be detected, and larger audit committees are associated with a higher likelihood of detection.

These correlations are consistent with our rationale for including the control variables in the estimation.13

13 The estimation results also reveal some unexpected correlations for a number of control variables, some of which are easier to explain than others. The positive correlation between sales growth rates and frauds may reflect that some of the high sales growth rates were the results of misleading sales figures. The negative coefficients on fraud and positive coefficients on detection for CEO-chair and CEO tenure may simply reflect selection: CEOs with better performance tend to chair the board and have longer tenure, and better performing firms are less likely to commit fraud. Those CEOs are also more closely watched and monitored,

20

III. EXPECTED COSTS OF COMMITTING FRAUD

Thus far, we have shown CEO connectedness significantly increases the incidence of fraud and

decreases the likelihood of detection. In this section we provide further collaborating evidence by

investigating possible channels through which CEO connectedness reduces the expected cost of

committing frauds. We assume when people decide whether to commit fraud, they weigh benefits of

wrongdoing against the expected cost, which is the probability of detection times the penalties. We focus

only on the cost side of the equation, because we do not have specific predictions on how CEO

connectedness directly affects the benefits of fraudulent activities.

III.1 Forced CEO Turnover-Fraud Sensitivity

There are two types of penalties CEOs may receive when their firms are detected with frauds

involving top management: court determined penalties, such as civil monetary and criminal penalties (e.g.,

jail); and/or market determined penalties, such as reputational loss and dismissals (Khanna, 1996). Unlike

court determined fines or jail terms, CEO dismissal is largely a firm level decision, which can be affected

by CEO connectedness within executive suites; for example, connected executives may support the

retention of a CEO tainted by fraud. In this section we relate CEO connectedness to the likelihood of

CEO dismissal given fraud by estimating the forced CEO turnover-fraud sensitivity.

To identify forced CEO turnover, we follow Parrino (1997) and Jenter and Kanaan (2010). If a

CEO’s departure is reported by the press as fired, forced out, or retirement or resignation due to policy

differences or pressure, it is classified as forced. All other departures for CEOs of 60 years or older are

classified as voluntary, unless they resign due to litigation or fraud. All departures for CEOs under 60 years

of age are evaluated further and are classified as forced if the press does not report the reason as death, poor

health, or acceptance of another position (including the chairmanship of the board); or if the article reports

resulting in a higher likelihood of detection. A similar selection explanation may also apply for the percent of non-independent directors in audit committee. Finally, the coefficients for industry litigation suggest that firms belonging to industries under intense litigation are associated with a higher likelihood of fraud and a lower probability of detection. This counter-intuitive result may be due to the substantial time lag between fraud incidence and detection (the average duration is 760 days). If the litigation intensity at the industry level is mean-reverting because regulators tend to focus on a particular industry one at a time, the current litigation intensity is negatively correlated to the future litigation intensity of fraud taking place now.

21

that the CEO is retiring, but does not announce the retirement at least six months before the succeeding

CEO takes office. Finally, cases classified as forced are reclassified as voluntary if press reports

convincingly explain the departure is due to previously undisclosed personal or business reasons that are

unrelated to the firm’s activities.

The estimation results are reported in Table VII. The dependent variable is Forced_CEO_Turnoverit,

an indicator for forced CEO turnover. The variable relating fraud to forced turnover of CEO in year t is

Fraud_t-2-t, which is equal to one if fraud takes place anytime during the three-year period over year t-2 to

year t. We use the three-year period to allow time for the fraud to be detected and for the CEO to be

punished. The results are robust to using a two-year period (unreported).14 The variables of main interests

are our measures of CEO connectedness and their interaction with Fraud_t-2-t. The interaction term

measures how CEO connectedness affects the CEO dismissal-fraud sensitivity.

We include all control variables in the fraud regression in Table V to avoid the potential omitted

variable problem. We add an indicator, Jail, to control for whether a CEO goes to jail. If a CEO is

sentenced to jail, he is unlikely to avoid dismissal regardless of how closely he is connected. Since

Forced_CEO_Turnoverit is an indicator variable, we use two alternative estimation methods: the OLS with

firm- and year-fixed effects in Columns (1) and (2), and the conditional logit model controlling for year

dummies in Columns (3) and (4). Robust standard errors are clustered at the firm level.

As expected, fraud is significantly and positively related to the likelihood of forced CEO turnover.

The results also show forced CEO turnover is less likely when CEOs are more connected in executive

suites. Perhaps most interesting, the positive forced CEO turnover-fraud sensitivity is reduced by CEO

connectedness; Fraud_t-2-t interaction with FTA or AFTA shows a significantly negative coefficient under

14 An alternative would be to relate fraud detection to forced CEO turnover. We do not follow this approach because by the time a fraud is detected the CEO involved in the fraud already may be replaced by a new CEO unassociated with the fraud. The median time lag between the commencement of fraud and its detection for our private litigation sample is 1.7 years (622 days), while the median CEO tenure is 4 years. So if we assume a fraud commences in the middle of a CEO’s tenure, there is a good chance the CEO liable for the fraud is replaced by a new CEO by the time the fraud is detected. Furthermore, this approach would exclude cases covered only by the SEC litigation releases because they provide insufficient information to accurately determine when frauds are detected.

22

all four specifications. Column (3) indicates that the probability of forced CEO turnover following a fraud

is 29.6% lower for a firm with all top four executives appointed during the CEO tenure (FTA = 1) than

for a firm with no top four executive appointed during the CEO tenure (FTA = 0.)

To address potential endogeneity, we again estimate two-stage IV regressions, where the IVs are

Num_Death and Num_Retire. The results are reported in Table A3 in the Appendix. FTA, AFTA, Fraud_t-

2-t *FTA, and Fraud_t-2-t*AFTA are endogenous in this analysis. The first-stage estimation results are

reported in Columns (1)-(2) and (4)-(5). The second stage IV regression estimation results are reported in

Columns (3) and (6). The predicted Fraud_t-2-t *FTA has a significant, negative coefficient. The predicted

Fraud_t-2-t *AFTA also shows a negative, albeit insignificant, coefficient. Based on these results, we

conclude CEO connectedness reduces the likelihood of fraud leading to a CEO dismissal.

III.2 Fraud Detection Duration

CEO connectedness may also reduce the expected costs of wrongdoing by hindering the detection

process. Top executives are often in a position both to receive internal information about wrongdoing and

to do something to interdict it (Dyck et al., 2010; Bowen et al., 2010).15 If the executives owe their

current positions to the CEO, they might be less enthusiastic about revealing information about

wrongdoing. Favors can go the other way as well. When a connected executive commits wrong doing, the

CEO may be more forgiving and help conceal the fraud.

If connectedness hinders fraud detection, the more connected a CEO is within executive suites,

the lower will be the probability of detection, and the longer it would take to detect a fraud. Thus, we

relate CEO connectedness to fraud detection duration, using cross-sectional data for each fraud case. The

dependent variable is the logged value of the number of days from the commencement of fraudulent

activity to the filing of class action litigation. The independent variables are the average FTA, AFTA, and

control variables over each fraud duration period. This test is based only on the private litigation data

15 Dyck et al., (2010) find that access to information is very important for fraud detection. Having access to inside information rather than relying on only public information increases the probability of detecting fraud by 15 percentage points.

23

from the SSCAC, because as mentioned earlier, the SEC litigation database does not provide sufficient

information to estimate the fraud detection duration.

The control variables include Tobin’s Q, leverage, sales growth rate, firm size, stock volatilities,

industry litigation risk, analyst coverage, CEO age, and industry dummies. Data requirements reduce the

sample to 246 unique fraud cases from the original 282 fraud cases. To avoid further reduction in the

sample size due to missing variables, we do not include board- and other CEO related variables and IOC,

which are missing more often than the above variables. (The next section focuses exclusively on the

board variables and IOC.) To account for possible correlations among fraud cases in the same industry,

robust standard errors are clustered at the industry level. The results are robust to clustering at the year of

the commencement of fraudulent activity.

Table VIII, Panel A, Columns (1) and (2) contain the estimation results, which show the number

of days frauds remain undetected is positively and significantly related to FTA and AFTA. The estimated

coefficient of FTA in the regression suggests that a fraud conducted by a firm with all top four executives

appointed during the CEO tenure (FTA = 1) will take 348 days longer to be detected than a fraud by a

firm with FTA = 0. In panel B, the dependent variable is the hazard ratio for the Cox regression, the

probability of detection in the next unit of time. Consistent with the OLS estimate, Columns (3) and (4)

show the hazard ratio is significantly and negatively related to both measures of CEO connectedness.

To address endogeneity, we again estimate two-stage IV regressions for fraud detection duration,

where the IVs are average Num_Death and Num_Retire during the fraud duration.16 The results, reported

in Table A4 in the Appendix, are robust.

III.3 Coordination Costs and the Number of Executives Charged

CEO connectedness can also reduce the cost of fraud when wrongdoing requires coordinated

action among multiple players. When CEOs are more connected with their top executives, the

16 For this regression analysis, we lower the retirement age to 65 from 70 because this sample of fraud firms shows no permanent retirement at the age of 70 or older.

24

environment in executive suites becomes more conducive for coordinated activities, including fraud. With

lower coordination costs, it is less costly to engage in frauds requiring cooperation from a number of

executives. Hence, we might witness more top executives being involved with, and charged in,

wrongdoing by firms with high CEO connectedness.

To test this prediction, we relate the number of executives named in a suit to FTA and AFTA in

Table IX. The dependent variable, Num_Charged, takes the value of one if the number of executives

charged in a litigation is less than or equal to two (N 2); two if the number of executives charged is

between three and five (3 N 5); and three if the number of executives charged is greater than five (N >

5). We rely on these indicators because when fraud requires coordinated action (or inaction) on the part of

executives, the likelihood of fraud may not rise linearly with the number of executives charged. For

example, if a fraud needs acquiescence from a majority of top executives, then we would expect non-

linearity around the majority threshold. An analogous approach is used in Ferris et al. (2003) when

examining the impact of multiple directorships on monitoring. Table III shows in a typical fraud litigation,

three to five executives are charged.

The set of control variables for this analysis differs sharply from previous regressions, because

the purpose is to control for factors affecting the number of executives charged. Some types of

misbehavior may require more coordination than others, leading to more executives being involved. For

example, inflating earnings requires a number of people to agree to the earnings figures (or at least not

oppose them), such as the CEO, CFO, accountant, lawyers and so forth, compared to insider trading

which may require only one person. Thus, we include three dummy variables indicating three types of

frauds: Accounting, Operating, and Executive. Accounting frauds are defined as misleading information

about financial condition, expected growth, financial statements; misleading information to inflate stock

price; violation of GAAP; and having to restate past financial statements. Operating frauds include cases

related to real corporate business activities; for example, a pharmaceutical company not disclosing

dangerous side effects; a company violating environmental regulations; and a bank misleading customers.

25

Executive frauds are defined as executives taking unlawful advantage of their positions to profit

themselves; for example, insider trading, related-party transactions, and so on. A fraud case may belong

to multiple types.

We also control for Tobin’s Q, sales growth, leverage, and industry dummies. If bad business

conditions associated with fraud manifest in the form of lower firm valuation, slow sales growth, high

financial leverage, worse business conditions may be associated with more serious frauds, which might

require cooperation from more executives. Industry dummies are included because the nature of frauds

varies across industries. We account for possible correlation among frauds in the same industry by

clustering standard errors at the industry level.

The estimation is again based on cross-sectional data. Because this analysis includes fewer

control variables than the fraud detection duration analysis, there are fewer missing variables, allowing us

to use 275 unique fraud cases. As before, the CEO connectedness variables and the control variables are

their averages over each case’s fraud duration.

Table IX reports estimation results, which show the number of executives charged with fraud is

positively and significantly related to FTA and AFTA. IV regression results are reported in Table A5 in

the Appendix. The first-stage results are reported in Column (1) and (3). We use only Num_Retire as an

IV.17 Num_Death is not included, because it is less likely that a dead person would be charged.18 When

dead people are not charged, the number of deaths will be negatively correlated to the number of

executives charged. The predicted values of FTA and AFTA in the second-stage estimation are both

positively and significantly related to the number of executives charged in frauds.19

III.4. Additional Robustness Tests

Organizational Changes

17 As in the fraud detection duration regressions, we lower the retirement age to 65 from 70 because this sub-sample also shows no permanent retirement at the age of 70 or older. 18 Technically, a dead person can be sued through her estate, but probably it occurs far less often than suits against living people. 19 Unlike other IV regressions, the first-stage estimation indicates the IV is weak and, thus, we cannot infer a causal relation based on these results.

26

Our results may be driven by major structural changes accompanying mergers, acquisitions,

divestitures, spinoffs, and other activities affecting organizational structure, which may change the

composition of executive suites, affecting our measures of CEO connectedness. The structural changes

may also affect incentives to commit fraud; for example, a bad acquisition and ensuing poor performance

may cause unlawful attempts to cover up losses. Thus, we add to the fraud regression the number of

mergers and acquisitions, MAit-1,and divestitures and spinoffs, DSit-1, completed in the previous year, and

Segmentsit, the number of business segments a firm has in a given year as reported in Compustat Segment

Data. These control variables reflect organizational changes due to acquisition, restructuring, and other

investment activities. Table A6 in the Appendix reports the re-estimation results of the bivariate models in

Table V. The results are robust.

Clustering Standard Errors at the Firm and the CEO-firm Pair Level

In Table V, robust standard errors are clustered at the industry level to account for possible

correlations among firms in the same industry. Because observations associated with a firm tend to be

correlated, we also cluster standard errors at the firm level and at the CEO-firm pair level. The re-

estimation results, reported in Table A7, are robust.

Alternative Measures of FTA and AFTA

In constructing FTA and AFTA, we treat all four top non-CEO executives equally. However, their

relative influence may differ and a CEO’s connectedness with, say, the number two executive may have

different ramifications to her internal influence than her connectedness with the number four executive.

Thus, we weight our measures of CEO connectedness by non-CEO executives’ salaries and bonuses.

FTA_W is calculated the same way as FTA, except that the fraction of top four non-CEO executives

appointed during the current CEO’s tenure is weighted by the sum of each non-CEO executive’s salaries

and bonuses. AFTA_W is the regression residual based on FTA_W.

The summary of statistics of FTA_W and AFTA_W are reported in Table III. The fraud sample

shows significantly higher average FTA_W and AFTA_W than the non-fraud sample. The pair-wise

27

correlations in Table IV, Panel A also show both weighted measures are significantly positively related to

the fraud indicator. We re-estimate the bivariate models in Table V with FTA_W and AFTA_W and report

the results in Table A8, Panel A. The results are robust. We also re-estimate the regressions for forced

CEO turnover-fraud sensitivity, fraud detection duration, and the number of executives charged with the

weighted CEO connectedness measures. The results are reported in Table A8, Panels B, C, and D,

respectively. All results are robust.

Industry Matched Sample

By construction, our sample contains many more non-fraud firms than fraud firms. Since some

non-fraud firms may not be comparable to the fraud firms, we re-estimate the bivariate model with an

industry-matched sample, which excludes all non-fraud firm year observations that do not have a fraud

firm-year observation in the same industry. The results are robust (unreported).

IV. INTERNAL AND EXTERNAL MONITORING MECHANISMS

Because deterring and detecting frauds is an important purpose of monitoring, a natural question

follows: Do monitoring mechanisms mitigate the adverse effects of CEO connectedness? In this section

we address this question by estimating the interactive effect of CEO connectedness with internal and

external monitoring mechanisms.

We proxy for the strength of internal monitoring by the board and audit committee composition:

the fraction of non-independent directors, %_NonIndepDirectors, and the fraction of non-independent

directors on the audit committee, %_NonIndepDirectors_Audit. Table X, Panel A repeats the bivariate

probit model estimation in Table V while interacting the proxies for internal monitoring with FTA or

AFTA in both fraud incidence and detection regressions. The estimated coefficients of FTA and AFTA

remain significant as before, but none of the interaction terms is significant. (Control variables are the

same as in Table V but are not reported.)

Panel B repeats the same exercise for external monitoring mechanisms, the strength of which is

proxied by institutional ownership concentration, IOC. Again, no interaction terms are significant. The

28

signs on IOC are in the right direction; negative for fraud incidence and positive for detection in all four

specifications; however, none of them are significant.

These results on internal and external monitoring suggest that the adverse influence of CEO

connectedness on corporate wrongdoing cannot be easily countered by improving the standard internal

and external monitoring measures, underscoring the importance of CEO connectedness as a factor in

assessing the risk of corporate fraud.

V. CONCLUSION

Allegations of corporate wrongdoing have peppered the headlines over the last decade capturing

the attention of the public, the business community, and regulatory agencies, while also damaging

investor confidence and shareholder value. This increased attention has led to an increase in scholarly

inquiry into how corporate wrongdoing arises and ways to deter and prevent it. The collective behavior of

corporate leaders is often critical in corporate wrongdoing, and the CEO often plays the central role. Yet

there are no studies exploring how CEOs’ influence within executive suites impact corporate wrongdoing.

This paper empirically examines the effects of CEO connectedness accumulated during the CEO’s tenure

through top executive appointment decisions.

We find the CEO connectedness is positively related to the likelihood of corporate fraud. The

relation is economically meaningful, statistically significant, and is robust to instrumental variables

regressions using top executive deaths and permanent retirements as IVs. It is also robust to major

organizational changes, clustering standard errors at different levels, alternate proxies for CEO

connectedness, and alternate sample constructions.

We also identify likely channels through which CEO connectedness influences wrongdoing—by

delaying detection, lowering the likelihood of CEO dismissal after fraud discovery, and reducing

coordination costs of conducting frauds, all of which reduce the expected costs of wrongdoing.

Furthermore, the adverse effects are not mitigated by standard measures of internal and external

monitoring mechanisms.

29

Taken together, these results imply that CEO connectedness (1) is a critical factor in assessing a

firm’s likelihood of engaging in wrongdoing, (2) has effects that are not mitigated by standard monitoring

mechanisms, and thus (3) is something to which investors, regulators, and governance specialists should

pay attention. Further, our results underscore the importance of network connections in the quality of

governance. Our focus on the connections between a CEO and his top executives helps identify how

connections built through personnel decisions may magnify the risk of corporate fraud.

30

References

Adams, Renee B., and Ferreira Daniel, 2007, A theory of friendly boards, Journal of Finance 62, 217-250.

Adams, Renee B., Heitor Almeida, and Daniel Ferreira, 2005, Powerful CEOs and their impact on corporate

performance, Review of Financial Studies 18, 1403-1432. Agrawal, Anup, and Sahiba Chadha, 2005, Corporate governance and accounting scandals, Journal of

Law and Economics 48, 371-406. Alexander, Cindy R., and Mark A. Cohen, 1999, Why do corporations become criminals? Ownership,

hidden actions, and crime as an agency cost, Journal of Corporate Finance 5, 1–34. Allen, William T., Reinier H. Kraakman and Guhan Subramanian, 2011, Commentaries and cases on the

law of business organization (Aspen PublishersWolters Kluwer). Arlen, Jennifer H., and William J. Carney, 1992, Vicarious liability for fraud on securities markets:

Theory and evidence, University of Illinois Law Review, 691 – 740. Bar-Gill, Oren and Lucian A. Bebchuk, 2002, Misreporting corporate performance. Harvard Law and

Economics Discussion Paper No. 400. Available at SSRN: http://ssrn.com/abstract=354141 or doi:10.2139/ssrn.354141.

Beasley, Mark S., 1996, An empirical analysis of the relation between the board of director composition

and financial statement fraud, Accounting Review 71, 443-465. Beatty, Randolph P., Howard Bunsis, and John R.M Hand, 1998, The indirect economic penalties in SEC

investigations of underwriters, Journal of Financial Economics 50, 151-186. Bebchuk, Lucian, and Jesse Fried, 2004, Pay without performance: The unfulfilled promise of executive

compensation (Harvard University Press). Bebchuk, Lucian, Martjin Cremers, and Urs Peyer, 2011, The CEO payslice, Journal of Financial

Economics 102, 199-221. Bennedsen, Morten, Francisco Perez-Gonzalez, and Daniel Wolfenzon, 2006, Do CEOs matter? NYU

Working Paper No. FIN-06-032. Available at SSRN: http://ssrn.com/abstract=1293659. Bertrand, Marianne, and Antoinette Schoar, 2003, Managing with style: The effect of managers on

corporate policy, Quarterly Journal of Economics 118, 1169-1208. Bhagat, Sanjai, John Bizjak, Jeffrey L. Coles, 1998, The shareholder wealth implications of corporate

lawsuits, Financial Management 27, 4 - 27. Black, Bernard S., Brian R. Cheffins, and Michael Klausner, 2006a, Outside director liability: A policy

analysis, Journal of Institutional and Theoretical Economics 162, 5-20.

31

Black, Bernard S., Brian R. Cheffins, and Michael Klausner, 2006b, Outside director liability, Stanford Law Review 58, 1055-1159.

Brav, Alon and Reuven Lehavy, 2003. An empirical analysis of analysts’ target prices: Short-term

informativeness and long-term dynamics. Journal of Finance 58, 1933-1967. Bowen, Robert, Andrew Call and Shiva Rajgopal, 2010, Whistle-blowing: Target firm characteristics and

economic consequences, Accounting Review 85, 1239-1271. Burns, Natasha, and Simi Kedia, 2006, The impact of performance-based compensation on misreporting,

Journal of Financial Economics 79, 35-67. Cialdini, Robert B., 1984, Principles of automatic influence. In J. Jacoby & S. Craig (Eds.), Personal

Influence: Theory, research, and practice (PP. 1-28). Lexington, MA: Heath. Chang, Xin, Sudipto Dasgupta, and Gilles Hilary, 2006, Analyst coverage and financing decisions,

Journal of Finance 61, 3009-3048. Choi, Stephen J., 2007, Do the merits matter less after the Private Securities Litigation Reform Act?

Journal of Law, Economics and Organization 23, 598-626. Choi, Stephen J., Karen K. Nelson, A.C. Pritchard, 2009, The screening effect of the Private

Securities Litigation Reform Act, Journal of Empirical Legal Studies 6, 35-68. Cohen, Lauren, Andrea Frazzini, and Christopher J. Malloy, 2008, The small world of investing: Board

connections and mutual fund returns, Journal of Political Economy 116, 951-979. Cohen, Lauren, Andrea Frazzini, and Christopher J. Malloy, 2010, Sell-side school ties, Journal of

Finance 65, 1409-1437. Cox, James D., Dana Kiku, and Randall S. Thomas, 2003, SEC enforcement heuristics: An empirical

inquiry, Duke Law Journal, 53, 737-779. Cronqvist, Henrik, Anil K. Makhija, and Scott E. Yonker, 2012, Behavioral consistency in corporate

finance: CEO personal and corporate leverage? Journal of Financial Economics 103, 20-40. Cremers, K.J. Martijn, and Vinay B. Nair, 2005, Governance mechanisms and equity prices, Journal of

Finance 60, 2859-2894. Das, Somnath, Re-Jin Guo, and Huai Zhang, 2006. Analysts’ selective coverage and subsequent

performance of newly public firms, Journal of Finance 61, 1159-1185. Del Guercio, Diane, Laura Seery, and Tracie Woidtke, 2008, Do boards pay attention when institutional

investor activists “Just Vote No”? Journal of Financial Economics 90, 84-103. Deli, Daniel N., and Stuart Gillan, 2000, On the demand for independent and active audit committees,

Journal of Corporate Finance 6, 427-45.

32

Dyck, I. J. Alexander, Adair Morse, and Luigi Zingales, 2010, Who blows the whistle on corporate fraud? Journal of Finance 65, 2213-2254.

Edmans, Alex, 2009, Blockholder trading, market efficiency, and managerial myopia, Journal of Finance

64, 2481-2514. Efendi, Jap, Anup Srivastava, and Edward Swanson, 2007, Why do corporate managers misstate financial

statements? The role of in-the-money options and other incentives, Journal of Financial Economics 85, 667-708.

Eisenberg, Theodore, Stefan Sundren, and Martin T. Wells, 1998, Larger board size and decreasing firm

value in small firms, Journal of Financial Economic 48, 35-54. Fama, Eugene F., and Kenneth R. French, 1997, Industry costs of equity, Journal of Financial Economics

43, 153-193. Ferris, Stephen P., Murali Jagannathan, and A. C. Pritchard, 2003, Too busy to mind the business?

Monitoring by directors with multiple board appointments, Journal of Finance 58, 1087-1112. Fracassi, Cesare, and Geoffrey Tate, 2011, External networking and internal firm governance, Journal of

Finance, forthcoming. Gande, Amar, and Craig M. Lewis, 2009, Shareholder-initiated class action lawsuits: Shareholder wealth

effects and industry spillovers. Journal of Financial and Quantitative Analysis 44, 823-850. Giroud, Xavier, and Holger M. Mueller, 2010, Does corporate governance matter in competitive

industries? Journal of Financial Economics 95, 312-331. Grundfest, Joseph A., 1995, Why disimply? Harvard Law Review 108, 740-741. Guadalupe, Maria, and Julie Wulf, 2010, The flattening firm and product market competition: The effect of

trade liberalization, American Economic Journal: Applied Economics 2, 105-127. Hartzell, Jay, and Laura T. Starks, 2003, Institutional investors and executive compensation, Journal of

Finance 58, 2351-2374. Hertzberg, Andrew, 2005, Managerial incentives, misreporting, and the timing of social learning: A

theory of slow booms and rapid recessions, Northwestern University Working Paper. Hochberg, Yael, Alexander Ljungqvist, and Yang Lu, 2007, Whom you know matters: Venture capital

networks and investment performance, Journal of Finance 62, 251-301. Hochberg, Yael, Alexander Ljungqvist, and Yang Lu, 2010, Networking as a barrier to entry and the

competitive supply of venture capital, Journal of Finance 65, 829-859. Hong, Harrison, Terence Lim, and Jeremy C. Stein, 2000. Bad news travels slowly: Size, analyst

coverage, and the profitability of momentum strategies. Journal of Finance 55, 265-295.

33

Hwang, Byoung-Hyoun, and Seoyoung Kim, 2009, It pays to have friends, Journal of Financial Economics 93, 138-158.

Michael C. Jensen, 1993, The modern industrial revolution, exit, and the failure of internal control

systems, Journal of Finance 48, 831-880. Jenter, Dirk, and Katharina Lewellen, 2011, CEO preference and acquisitions, Stanford University

Working Paper. Jenter, Dirk and Fadi Kanaan, 2011, CEO turnover and relative performance evaluation, Journal of

Finance, forthcoming. Johnson, Marilyn F., Karen K. Nelson, and Adam C. Pritchard, 2007, Do the merits matter more? The

impact of the Private Securities Litigation Reform Act, Journal of Law, Economics, and Organization 23, 627-652.

Jones, Christopher, and Seth Weingram, 1996, The determinants of 10b-5 litigation risk, Stanford Law

School Working Paper. Karpoff, Jonathan M. and John R. Lott, Jr., 1993, The reputational penalty firms bear from committing

criminal fraud, Journal of Law and Economics 36, 757-802. Karpoff, Jonathan M., D. Scott Lee, and Valaria P. Vendrzyk, 1999, Defense procurement fraud, penalties,

and contractor influence, Journal of Political Economy 107, 809-842. Karpoff, Jonathan M., D. Scott Lee, and Gerald S. Martin, 2008a, The cost to firms of cooking the books,

Journal of Financial and Quantitative Analysis 43, 581-611.

Karpoff, Jonathan M., D. Scott Lee, and Gerald S. Martin, 2008b, The consequences to managers for cooking the books, Journal of Financial Economics 88, 193-215.

Karpoff, Jonathan M., D. Scott Lee, and Gerald S. Martin, 2010, Bribery: Business as usual? Available at SSRN: http://ssrn.com/abstract=1573222. Kelly, Bryan and Alexander Ljungqvist, 2011, Testing asymmetric-information asset pricing models.

Review of Financial Studies, forthcoming. Khanna, Vikramaditya, 2003, Should the behavior of top management matter? Georgetown Law Journal

91, 1215 – 1256. Khanna, Vikramaditya S. 1996, Corporate criminal liability: What purpose does it serve?, Harvard Law

Review 109, 1477 – 1534. Kim, E. Han, and Yao Lu, 2011, CEO ownership, external governance, and risk-taking, Journal of

Financial Economics 102, 272-292. Kim, E. Han, and Yao Lu, 2012, The independent board requirement and CEO connectedness. Available

at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2019199.

34

Landier, Augustin, David Sraer, and David Thesmar, 2008, Bottom-up corporate governance, New York University Working Paper.

Larcker, David F., Scott A. Richardson, Andrew Seary, and A. Irem Tuna, 2005, Back door links between

directors and executive compensation. Available at SSRN: http://ssrn.com/abstract=671063. Lipton, Martin, and Jay W. Lorsch, 1992, A modest proposal for improved corporate governance,

Business Lawyer 48, 59-77. Morse, Adair, Vikram Nanda, and Amit Seru, 2011, Are incentive contracts rigged by powerful CEOs?

Journal of Finance 66, 1779–1821. Murphy, Deborah L., Ronald E. Shrieves, and Samuel L. Tibbs, 2009, Determinants of the stock price

reaction to allegations of corporate misconduct: Earnings, risk, and firm size effects. Journal of Financial and Quantitative Analysis 43, 581-612.

Parrino, Robert, 1997, CEO turnover and outside succession: A cross-sectional analysis, Journal of

Financial Economics, 46, p. 165-197. Peng, Lin, and Ailsa Röell, 2008, Executive pay and shareholder litigation, Review of Finance 12, 141-

184. Povel, Paul, Rajdeep Singh, and Andrew Winton, 2007, Booms, busts, and fraud, Review of Financial

Studies 20, 1219-1254. Shivdasani, Anil, and David Yermack, 1999, CEO involvement in the selection of new board members:

An empirical analysis, Journal of Finance 54, 1829-1853. Stuart, Toby E., and Soojin Yim, 2010, Board interlocks and the propensity to be targeted in private

equity transactions, Journal of Financial Economics 97, 174-189. Vafeas, Nikos, 1999, Board meeting frequency and firm performance, Journal of Financial Economics 53,

113-1142. Wang, Tracy Y. 2011, Corporate securities fraud: Insights from a new empirical framework, Journal of

Law, Economics and Organization, forthcoming. Wang, Tracy Y., Andrew Winton, and Xiaoyun Yu, 2010, Corporate fraud and business conditions:

Evidence from IPOs, Journal of Finance 65, 2255-2292. Weisbach, Michael S., 1988, Outside directors and CEO turnover, Journal of Financial Economics 20,

431-460. Yermack, David L., 1996, Higher market valuation of companies with a small board of directors, Journal

of Financial Economics 40, 185-212. Yu, Frank, and Xiaoyun Yu, 2011, Corporate lobbying and fraud detection, Journal of Financial and

Quantitative Analysis, forthcoming.

!"#

#

#

Table I . Variable Descr iptions and Data Sources.

Variables Definitions Sources Panel A : F raud Variables

Fraud An indicator equal to one if a firm-year observation shows a fraud record, and zero otherwise.

Stanford Securities Class Action Clearinghouse (SSCAC) and the SEC

Fraud_t-2-t An indicator equal to one if fraud takes place anytime during the three-year period over year t-2 to year t, and zero otherwise. SSCAC and the

SEC

Duration The number of days from the commencement of fraudulent activity to the filing of the class action litigation. SSCAC

Num_Charged

A variable equal to one if the number of executives charged in the litigation is less than or equal to between three and five ( 5); and three if the number of executives charged is greater than five (N >5).

SSCAC and the SEC

Accounting Indicator equal to one, if a fraud is identified as involving accounting matters. SSCAC and the SEC

Operating Indicator equal to one, if a fraud is identified as involving real business activities. SSCAC and the SEC

Executive Indicator equal to one if a fraud is identified as involving executives taking unlawful advantage of their positions for personal benefits. SSCAC and the

SEC

Jail Indicator equal to one if a CEO is sentenced to jail, and zero otherwise. SSCAC and the SEC

Panel B : C E O Connectedness Variables FTA Fraction of top four non- ExecuComp

AFTA Abnormal fraction of top four non-CEO executives appointed during the current ExecuComp

FTA_W Fraction of top four non- ExecuComp

AFTA_W Abnormal fraction of top four non-CEO executives appointed during the current ExecuComp

Panel C : Variables to Construct A F TA CEO_Tenure The number of years a CEO has been in office. ExecuComp Outside Indicator equal to one, if a CEO comes from outside the firm and zero otherwise. ExecuComp Execsen The average number of years four top non-CEO executives have been in office. ExecuComp

FTA_1Y The fraction of top four non- year in office. ExecuComp

Unknown The fraction of executives whose first year on the list of top four non-CEO executives cannot be identified. ExecuComp

FTA_1Y_Unknown The fraction of top four non-CEO executives whose appointment cannot be determined as occurring ExecuComp

Panel D: F irm Business Condition Variables

Tobin's Q The market value of common equity plus the book value of total liabilities divided by the book value of total assets. Compustat

Ebitda/TA Earnings before interest, taxes, depreciation, and amortization divided by the book value of total assets. Compustat

Leverage Sum of short- and long-term debt divided by the book value of total assets. Compustat SalesGrowth_3Yr The 3-year least squares annual growth rate of sales in percentage. ExecuComp Ln(TotalAssets) Logged value of the book value of total assets. Compustat

IndustryQ The median Tobin's Q in an industry in a given year. Industries are defined by Fama-French (1997) industry groupings. Compustat

ICR Industry concentration ratio, as measured by the sum of the percentage market share (in sales) of the four biggest firms among all firms in Compustat in each industry in each year as defined by Fama-French (1997).

Compustat

!$#

#

##

Table I . Variable Descr iptions and Data Sources (Continued). Variables Definitions Sources Panel E : Corporate Governance Variables Ln(BoardSize) Logged value of one plus the number of directors on the board. IRRC/ Risk Metrics

%_NonIndepDirectors The number of non-independent directors, as defined by IRRC, divided by the total number of directors on the board. IRRC/ Risk Metrics

Ln(BoardMeetings) Logged value of one plus the number of board meetings held during a given year. ExecuComp

%_NonIndepDirectors_Audit The number of non-independent directors as defined by IRRC on the audit committee, divided by the total number of audit committee members. IRRC/ Risk Metrics

Ln(AuditComSize) Logged value of one plus the number of audit committee members. IRRC/ Risk Metrics

Ln(Analyst) Logged value of one plus the number of analysts following a firm in a given year. I/B/E/S

IOC The sum of percentage share ownership held by the top five institutional investors.

CDA Spectrum Database

Panel F : Litigation Risk Variables StockVolatilities Standard deviation of daily stock returns in a given year. CRSP

IndustryLitigation The number of litigations against publicly-listed firms in an industry in a given year, divided by the number of firms in Compustat for the same industry and the same year. Industries are defined by Fama-French (1997) industry groupings.

SSCAC, the SEC, and Compustat

Panel G: C E O Characteristics Variables CEO_OWN The percentage of outstanding common shares held by a CEO. ExecuComp

CEO_Founder An indicator equal to one, if a CEO was the CEO five years prior to the first date the firm appears in CRSP or Compustat. ExecuComp

CEO_Chair Indicator for CEO also chairing the board and zero otherwise. ExecuComp Ln(CEO_Age) Logged value of CEO age. ExecuComp Panel H : Instrumental Variables for C E O Connectedness

Num_Death

The number of top four non-CEO executives who left the position due to death For inclusion, the

identified reason for leaving the company is death and the identified year of leaving the firm is the same as or one year after the last year in which the

in ExecuComp.

ExecuComp and Factiva

Num_Retire

The number of top four non-CEO executives who left the position due to

For inclusion, (i) an executive is 70 years or older at the time of retirement; (ii)

the ExecuComp; (iii) the identified reason for leaving the company is retirement; and (iv) the identified year of leaving the firm is the same as, or one year after, the last year in which the ex n ExecuComp.

ExecuComp and Factiva

Panel I : O ther Variables

Forced_CEO_Turnover

Indicator for forced CEO turnover, identified by following the procedures used in Parrino (1997) and Jenter and Kanaan (2010). If a CEO departure is reported by the press as the CEO is fired, forced out, or retires or resigns due to policy differences or pressure, it will be classified as forced. All other departures for CEOs above and including age 60 are classified as voluntary (except for the cases due to litigation or other fraud). All departures for CEOs below age 60 are evaluated further and are classified as forced either if the article does not report the reason as death, poor health, or the acceptance of another position (including the chairmanship of the board); or if the article reports that the CEO is retiring, but does not announce the retirement at least six months before the succession. Finally, cases classified as forced are reclassified as voluntary if press reports convincingly explain the departure as due to previously undisclosed personal or business reasons

ExecuComp and Factiva

MA The number of mergers and acquisitions completed by a firm in the previous year. SDC

DS The number of divestitures and spinoffs completed by a firm in the previous year. SDC

Segment The number of business segments a firm has in a given year. Compustat Segment Data

!%#

#

####

Table I I . Sample Distr ibution by Year and by F raction of Executives Appointed

This table describes the sample firm-year observations. Panel A lists the sample distribution by year; Panel B lists the sample distribution by the fraction of executives appointed (FTA) (1) shows the total number of firms with data available to calculate FTA. Columns (2) and (3) report the number and the percentage of firms committing fraud among the sample firms. Panel A : Sample Distribution by Year

(1) (2) (3)

Year # of F irms # of F irms with

F rauds %_F raud 1996 1,518 22 1.449 1997 1,551 41 2.643 1998 1,607 39 2.427 1999 1,684 67 3.979 2000 1,670 82 4.910 2001 1,562 91 5.826 2002 1,574 86 5.464 2003 1,630 62 3.804 2004 1,637 60 3.665 2005 1,628 67 4.115 2006 1,736 67 3.859 Total 17,797 684 3.843

Panel B : Sample Distribution by F TA

F T A # of F irms # of F irms with

F rauds %_F raud 0.00 4,988 163 3.268 0.25 3,586 112 3.123 0.50 4,033 153 3.794 0.75 3,268 146 4.468 1.00 1,922 110 5.723

#

!&#

#

#Table I I I . Summary Statistics of K ey Variables for the Full, F raud, and Non-F raud Samples. This table reports summary statistics for key variables. Panel A contains the statistics for the full sample. Panel B reports the mean of each variable separately for the fraud and non-fraud sample. Columns (8) and (9) show for each variable the difference in mean between the fraud and non-fraud sample and the P-value of the difference, respectively. The definitions of all variables are provided in Table I.

Panel A : Full Sample Panel B : F raud and Non-F raud F irm

Samples (1) (2) (3) (4) (5) (6) (7) (8) (9)

Variable M ean M edian Std. Dev. M in M ax F raud

Sample

Non-F raud

Sample

F raud Sample - Non-F raud

Sample P-value F raud Variables Fraud 0.038 0.000 0.192 0.000 1.000 Accounting 0.720 1.000 0.450 0.000 1.000 Operating 0.273 0.000 0.446 0.000 1.000 Executive 0.248 0.000 0.433 0.000 1.000 Duration 760.342 622.000 537.195 13.000 2301.000 Fraud_t-2-t 0.049 0.000 0.215 0.000 1.000 Num_Charged 1.947 2.000 0.720 1.000 3.000 Jail 0.001 0.000 0.024 0.000 1.000 C E O Connectedness Variables FTA 0.409 0.500 0.336 0.000 1.000 0.474 0.407 0.067 (0.000) AFTA 0.000 0.005 0.307 -1.202 0.738 0.032 -0.002 0.034 (0.005) FTA_W 0.389 0.374 0.336 0.000 1.000 0.462 0.386 0.076 (0.000) AFTA_W -0.001 -0.063 0.306 -1.172 0.757 0.041 -0.002 0.043 (0.000) Business Condition Variables Tobin's Q 2.097 1.503 2.604 0.298 105.090 2.752 2.071 0.681 (0.000) Ebitda/TA 0.129 0.127 0.123 -2.948 0.991 0.110 0.130 -0.020 (0.000) Leverage 0.223 0.214 0.175 0.000 0.959 0.242 0.222 0.020 (0.005) SalesGrowth_3Yr 17.325 9.929 63.582 -91.136 3559.292 32.495 16.722 15.773 (0.000) Log(TotalAssets) 0.639 0.461 1.742 -5.419 7.541 1.433 0.607 0.826 (0.000) IndustryQ 1.509 1.344 0.491 0.842 3.497 1.585 1.505 0.080 (0.000) ICR 0.326 0.293 0.143 0.087 0.981 0.322 0.327 -0.005 (0.400) C E O Characteristics Variables CEO_OWN 0.025 0.003 0.062 0.000 0.761 0.024 0.025 -0.001 (0.561) CEO_Founder 0.137 0.000 0.344 0.000 1.000 0.181 0.135 0.046 (0.001) CEO_Chair 0.801 1.000 0.399 0.000 1.000 0.845 0.799 0.046 (0.019) Ln(CEO_Age) 3.313 3.367 0.290 0.000 4.159 3.234 3.316 -0.082 (0.000) CEO_Tenure 6.703 4.000 7.142 0.000 55.000 7.499 6.672 0.827 (0.003) Forced_CEO_Turnover 0.029 0.000 0.169 0.000 1.000 0.038 0.029 0.009 (0.173) Governance and Monitoring Variables Ln(BoardSize) 2.350 2.303 0.267 0.693 3.689 2.389 2.349 0.040 (0.002) %_NonIndepDirectors 0.336 0.308 0.174 0.000 1.000 0.324 0.337 -0.013 (0.134) Ln(BoardMeetings) 2.070 2.079 0.351 0.000 4.220 2.177 2.065 0.112 (0.000) %_NonIndepDirectors_Audit 0.033 0.000 0.068 0.000 0.778 0.033 0.033 0.000 (0.932) Ln(AuditComSize) 1.262 1.386 0.604 0.000 2.485 1.386 1.257 0.129 (0.000) Ln(Analyst) 1.966 2.197 1.059 0.000 3.951 2.319 1.952 0.367 (0.000) IOC 0.413 0.386 0.139 0.162 0.972 0.387 0.414 -0.027 (0.000) Litigation Risk Variables StockVolatilities 0.028 0.024 0.016 0.000 0.231 0.034 0.028 0.006 (0.000) IndustryLitigation 0.019 0.015 0.021 0.000 0.140 0.026 0.019 0.008 (0.000) Instrumental Variables for C E O Connectedness Variables Num_Death 0.019 0.000 0.167 0.000 2.000 0.004 0.020 -0.016 (0.016) Num_Retire 0.006 0.000 0.094 0.000 2.000 0.006 0.006 0.000 (0.899)

#

!'#

#

#Table I V . Pair-wise Cor relations between F raud Variables and C E O Connectedness. This table reports the pair-wise correlations among fraud variables and CEO connectedness variables. Panel A is based on panel data of all firm-year observations, reporting the pair-wise correlations between the fraud indicator variable and CEO connectedness variables. Panel B is based on cross-sectional data on fraud cases, reporting the pair-wise correlations among the CEO connectedness variables, fraud detection duration, the number of executives charged in litigation, and the type of fraud. The definitions of all variables are provided in the Table I. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. Panel A : F irm-Level Data Fraud FTA AFTA FTA_W AFTA_W Fraud 1 FTA 0.0382*** 1 AFTA 0.0212*** 0.9193*** 1 FTA_W 0.0437*** 0.9849*** 0.9021*** 1 AFTA_W 0.0269*** 0.9029*** 0.9821*** 0.9183*** 1

##

Panel B : Case-Level Data Duration Num_Charged Accounting Operating Executive FTA AFTA FTA_W AFTA_W Duration 1 Num_Charged 0.2508*** 1 Accounting -0.0587 0.0262 1 Operating -0.0709 0.1091* -0.4507*** 1 Executive 0.1477** 0.0317 -0.1534*** -0.1495** 1 FTA 0.1611*** 0.1169* -0.0313 0.0166 0.0653 1 AFTA 0.0791 0.1015* -0.0197 0.0421 0.0834 0.9258*** 1 FTA_W 0.1325** 0.1156* -0.0211 0.0271 0.0436 0.9894*** 0.9134*** 1 AFTA_W 0.0500 0.1020* -0.0089 0.0539 0.0608 0.9163*** 0.9882*** 0.9262*** 1

#

()#

#

#

Table V . Bivariate Model Estimation Results for C E O Connectedness and Corporate F rauds.

This table reports the bivariate model estimation results. Columns (1) and (3) report the relation between CEO connectedness and the incidence of fraud, and Columns (2) and (4) report the estimation results for detection of fraud, given fraud. Definitions of all variables are provided in Table I. All regressions include year dummies. Robust standard errors clustered at the industry level are reported in parentheses. Industries are classified by Fama-French 48 industry groupings. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively.

Model 1 Model 2 F raud Detect|F raud F raud Detect|F raud (1) (2) (3) (4) FTA 1.023*** -0.737** (0.294) (0.365) AFTA 1.273*** -0.949** (0.325) (0.382) Tobin's Q -0.282*** 0.193** -0.297*** 0.206** (0.093) (0.094) (0.092) (0.082) Ebitda/TA -2.377 1.238 -2.150 1.094 (1.495) (0.846) (1.493) (0.853) Leverage -1.421 1.354 -1.504 1.420 (1.967) (1.650) (1.651) (1.379) SalesGrowth_3Yr 0.014** -0.003 0.014** -0.004 (0.007) (0.003) (0.007) (0.003) Log(TotalAssets) -0.063 0.176 -0.056 0.171 (0.225) (0.128) (0.208) (0.113) IndustryQ 0.089 0.102 (0.635) (0.586) (IndustryQ)2 0.046 0.043 (0.182) (0.164) ICR 0.351 0.331 (0.564) (0.510) CEO_OWN -0.554 -0.590 (1.467) (1.433) (CEO_OWN)2 -1.390 -0.993 (5.824) (5.569) CEO_Founder 0.291 -0.220 0.335 -0.263 (1.306) (0.871) (1.092) (0.726) CEO_Chair -0.441 0.299 -0.426 0.291 (0.405) (0.275) (0.357) (0.252) Ln(CEO_Age) 0.859 -0.880 0.852 -0.874* (0.723) (0.550) (0.650) (0.505) CEO_Tenure -0.014 0.019 -0.002 0.011 (0.040) (0.034) (0.032) (0.031) Ln(BoardSize) -2.434 1.940* -2.530* 2.017** (1.700) (1.042) (1.449) (0.924) %_NonIndepDirectors -0.878 0.958 -0.902 0.967 (1.011) (0.996) (0.983) (0.952) Ln(BoardMeetings) 0.048 0.107 0.068 0.091 (0.664) (0.557) (0.547) (0.471) %_NonIndepDirectors_Audit -1.440 0.392 -1.400 0.399 (2.717) (2.156) (2.599) (2.006) Ln(AuditComSize) 0.602 -0.271 0.509 -0.203 (0.757) (0.526) (0.697) (0.475) Ln(Analyst) -0.172 0.192 -0.210 0.218 (0.354) (0.226) (0.302) (0.189) IOC -1.676 1.203 -1.746 1.258 (1.872) (1.810) (1.402) (1.394) StockVolatilities -20.536** 29.987*** -18.835* 28.548*** (10.108) (9.223) (9.660) (8.591) IndustryLitigation 24.253 -9.335*** 22.876 -8.846*** (15.986) (3.574) (15.633) (2.976) Constant 4.716* -5.028*** 5.293** -5.432*** (2.589) (1.925) (2.498) (1.795) Year Dummies Y Y Y Y Observations 7871 7871 7871 7871 Wald Chi2 -1033.7673 -1031.9532 Prob> Chi2 0.00 0.00

##

(*#

#

##

Table V I . Instrumental Variable Regressions for C E O Connectedness and Corporate F raud.

This table reports the second-stage estimation results of IV regressions to address potential endogeneity in CEO connectedness. The endogenous variables are FTA in Model 1; AFTA in Model 2. The instrumental variables are Num_Death and Num_Retire. All regressions control for year dummies. The first stage regression estimation results are reported in Table A2 in the Appendix. Definitions of all variables are provided in Table I. Robust standard errors clustered at the industry level are reported in parentheses. Industries are classified by Fama-French 48 industry groupings. Coefficients marked with *, **, and *** are significant at 10%, 5%, and 1%, respectively.

Model 1 Model 2 F raud Detect|F raud F raud Detect|F raud (1) (2) (3) (4) FTA_Hat 36.035** -29.285*** (14.424) (8.122) AFTA_Hat 34.892*** -29.997*** (13.102) (9.892) Tobin's Q -0.214** 0.140** -0.242** 0.161** (0.098) (0.067) (0.100) (0.066) Ebitda/TA -4.886** 3.378*** -4.734** 3.350*** (2.060) (1.001) (2.005) (1.104) Leverage -4.073*** 3.423*** -4.452*** 3.794*** (0.901) (0.869) (0.963) (0.928) SalesGrowth_3Yr 0.023** -0.014*** 0.027*** -0.017*** (0.009) (0.004) (0.010) (0.005) Log(TotalAssets) -0.477*** 0.462*** -0.542*** 0.528*** (0.148) (0.107) (0.150) (0.121) IndustryQ 0.298 0.290 (0.524) (0.527) (IndustryQ)2 -0.025 -0.022 (0.099) (0.100) ICR 0.206 0.200 (0.275) (0.278) CEO_OWN 0.035 0.114 (1.393) (1.395) (CEO_OWN)2 -1.724 -1.682 (4.556) (4.567) CEO_Founder 6.754** -5.520*** 8.393** -7.282*** (3.005) (1.587) (3.388) (2.468) CEO_Chair -0.979* 0.779*** -0.752* 0.623** (0.529) (0.298) (0.445) (0.291) Ln(CEO_Age) -0.022 -0.187 0.180 -0.376 (0.517) (0.353) (0.522) (0.373) CEO_Tenure -0.696** 0.572*** -0.353** 0.311*** (0.290) (0.160) (0.139) (0.104) Ln(BoardSize) -1.422 1.179 -1.654 1.392 (1.130) (0.878) (1.132) (0.870) %_NonIndepDirectors 11.229** -8.960*** 8.783** -7.471*** (5.139) (2.674) (3.985) (2.824) Ln(BoardMeetings) -1.752** 1.548*** -1.204** 1.158** (0.786) (0.516) (0.544) (0.457) %_NonIndepDirectors_Audit -13.920** 10.681*** -12.137** 9.755** (6.236) (3.910) (5.507) (4.150) Ln(AuditComSize) -0.384 0.527** -0.611 0.767** (0.429) (0.269) (0.439) (0.331) Ln(Analyst) 0.010 0.057 -0.067 0.126 (0.239) (0.143) (0.225) (0.136) IOC -2.506** 1.837* -2.013* 1.419 (1.163) (1.085) (1.136) (1.044) StockVolatilities -64.392*** 64.297*** -34.189** 40.226*** (18.623) (11.541) (13.341) (10.480) IndustryLitigation 13.371* -3.542 18.740** -7.682*** (7.772) (2.287) (8.774) (2.093) Constant -1.188 0.106 6.145*** -5.829*** (3.993) (2.442) (2.181) (1.816) Year Dummies Y Y Y Y Observations 7871 7871 7871 7871 Wald Chi2 -1002.7162 -1002.9303 Prob> Chi2 0.00 0.00

(+#

#

#Table V I I . Forced C E O Turnover-F raud Sensitivity. This table estimates the impact of CEO connectedness on the forced CEO turnover-fraud sensitivity. The dependent variable is an indicator of CEO forced turnover. Fraud_t-2-t is an indicator equal to one if fraud takes place anytime during the three-year period over year t-2 to year t, and zero otherwise. The sample covers the period 1996 through 2006. The results are estimated by the OLS in Panel A and by the conditional logit model in Panel B. Regressions in Panel A control for firm and year fixed effects; Regressions in Panel B control for year dummies. Definitions of all variables are provided in Table I. Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. Forced_C E O_Turnover Panel A : O LS Panel B : C logit (1) (2) (3) (4) Fraud_t-2-t*FTA -0.226*** -2.686** (0.066) (1.345) Fraud_t-2-t*AFTA -0.276*** -3.864*** (0.075) (1.486) FTA -0.101*** -2.602*** (0.015) (0.539) AFTA -0.122*** -2.708*** (0.018) (0.592) Fraud_t-2-t 0.168*** 0.074*** 2.175*** 1.225** (0.046) (0.023) (0.655) (0.557) Tobin's Q -0.001 -0.001 0.020 0.019 (0.002) (0.002) (0.038) (0.038) Ebitda/TA -0.034 -0.033 -0.526 -0.556 (0.050) (0.049) (1.868) (1.842) Leverage -0.017 -0.018 0.052 0.102 (0.035) (0.035) (1.442) (1.465) SalesGrowth_3Yr 0.000 0.000 0.001 0.001 (0.000) (0.000) (0.002) (0.002) Log(TotalAssets) 0.006 0.007 -0.238 -0.231 (0.011) (0.011) (0.525) (0.524) IndustryQ -0.031 -0.029 -1.480 -1.474 (0.047) (0.047) (2.545) (2.560) (IndustryQ)2 0.007 0.006 0.362 0.364 (0.010) (0.010) (0.557) (0.559) ICR -0.019 -0.016 -7.109* -7.298* (0.078) (0.078) (4.186) (4.365) CEO_OWN 0.028 -0.001 19.455 19.369 (0.199) (0.201) (13.989) (13.897) (CEO_OWN)2 -0.253 -0.207 -100.435** -104.814** (0.301) (0.303) (44.557) (45.619) CEO_Founder 0.091*** 0.078*** 5.059*** 5.018*** (0.024) (0.024) (1.807) (1.850) CEO_Chair -0.002 -0.002 -0.219 -0.219 (0.010) (0.010) (0.308) (0.309) Ln(CEO_Age) 0.087*** 0.089*** 1.691** 1.726** (0.027) (0.028) (0.781) (0.809) CEO_Tenure -0.006*** -0.007*** -0.479*** -0.506*** (0.001) (0.001) (0.139) (0.138) Ln(BoardSize) -0.024 -0.021 -1.495 -1.264 (0.020) (0.020) (0.919) (0.906) %_NonIndepDirectors -0.017 -0.017 0.519 0.643 (0.033) (0.033) (1.428) (1.421) Ln(BoardMeetings) 0.042*** 0.041*** 1.096** 1.026** (0.012) (0.012) (0.455) (0.454) %_NonIndepDirectors_Audit -0.029 -0.029 -2.393 -2.390 (0.050) (0.050) (2.104) (2.130) Ln(AuditComSize) 0.000 -0.000 0.396 0.436 (0.013) (0.013) (0.559) (0.554) Ln(Analyst) -0.017* -0.016 -0.874** -0.846* (0.010) (0.010) (0.436) (0.437) IOC 0.075** 0.073** 3.455** 3.333** (0.034) (0.034) (1.560) (1.574) StockVolatilities 0.420 0.400 23.368 23.449 (0.431) (0.430) (17.831) (17.839) IndustryLitigation 0.289* 0.290* 6.882 6.675 (0.164) (0.164) (4.497) (4.516) Jail 0.133* 0.133** 5.044*** 5.471*** (0.074) (0.067) (1.788) (1.872) Constant -0.209* -0.255** (0.118) (0.119) Firm FE & Year FE Y Y N N Year Dummies N N Y Y Observations 7871 7871 1195 1195 Adj-R2 (Pseudo R2) 0.12 0.13 (0.45) (0.45)

(!#

#

#

Table V I I I . F raud Detection Duration and the Hazard Ratio. This table relates CEO connectedness to fraud detection duration and the hazard ratio. The sample covers 246 fraud cases over the period 1996 through 2006. Panels A and B report the results estimated by the OLS and the Cox regressions. The dependent variable in Panel A is the logged value of the number of days from the commencement of fraudulent activity to the filing of the class action litigation plus one; in Panel B, the hazard ratio for the Cox regression. All regressions control for industry dummies. CEO connectedness and control variables are their average values over the fraud period. Definitions of all variables are provided in Table I. Robust standard errors clustered at the industry level are reported in parenthesis. Industries are classified by Fama-French 48 industry groupings. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. Panel A : Ln(Duration+1) Panel B : _t (1) (2) (3) (4) FTA 0.458** -0.870*** (0.180) (0.236) AFTA 0.281* -0.522*** (0.152) (0.197) Log(TotalAssets) -0.067 -0.059 0.038 0.025 (0.047) (0.048) (0.058) (0.060) Tobin's Q 0.020 0.023* -0.035 -0.041 (0.013) (0.013) (0.042) (0.046) Leverage 0.872** 0.820** -1.141** -1.065** (0.337) (0.340) (0.553) (0.535) SalesGrowth_3Yr 0.001** 0.001** -0.001 -0.001 (0.000) (0.000) (0.001) (0.001) StockVolatilities -17.351*** -17.626*** 19.438** 19.171** (5.519) (5.496) (8.802) (8.980) IndustryLitigation 1.177 1.662 -8.421 -9.400 (4.053) (3.950) (6.085) (6.124) Ln(Analyst) -0.079 -0.074 0.178** 0.161* (0.077) (0.077) (0.086) (0.083) Ln(CEO_Age) -0.273 -0.229 0.299 0.191 (0.183) (0.179) (0.326) (0.326) Constant 7.092*** 7.203*** (0.574) (0.605) Industry Dummies Y Y Y Y Observations 246 246 246 246 Adj-R2 0.11 0.09 Wald Chi2 21222.83 43617.14

#

((#

#

#

Table I X . Number of Executives Charged in L itigation . This table estimates the relation between CEO connectedness and the number of executives charged in litigation. The dependent variable Ln(Num_Charged) is the logged value of Num_Charged plus one. Num_Charged takes the value of one, if the number of executives charged in the litigation is less than three (N < 3); two, if the number of executives charged is between three and five ( ); and three if the number of executives charged is greater than five (N > 5). The sample covers 275 fraud cases over the period 1996 through 2006. The CEO connectedness variables and the control variables are their average values over the fraud period. All regressions control for industry dummies. Definitions of all variables are provided in Table I. Robust standard errors clustered at the industry level are reported in parentheses. Industries are classified by Fama-French 48 industry groupings. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. Ln(Num_Charged+1) (1) (2) FTA 0.123*** (0.032) AFTA 0.114*** (0.039) Tobin's Q 0.004** 0.004** (0.002) (0.002) SalesGrowth_3Yr -0.000*** -0.000*** (0.000) (0.000) Leverage 0.217** 0.203** (0.085) (0.087) Accounting 0.059 0.056 (0.035) (0.035) Operating 0.067** 0.066* (0.033) (0.034) Executive 0.012 0.011 (0.039) (0.039) Constant 0.869*** 0.918*** (0.069) (0.064) Industry Dummies Y Y Observations 275 275 Adj-R2 0.14 0.13

#

("#

#

#

Table X . Interactive E ffects of C E O Connectedness and Internal and External Monitoring M echanisms on Corporate F rauds. This table reports the bivariate model estimation results on interactive effects of CEO connectedness and monitoring mechanisms on corporate frauds. Panels A and B report the results on the interactive effects for internal and external monitoring mechanisms, respectively. Columns (1), (3), (5) and (7) report the results of the fraud incidence regressions. Control variables are identical to those in Columns (1) and (3) of Table V but their coefficients are not reported. Columns (2), (4), (6) and (8) report the results of the fraud detection regressions. Control variables are identical to those in Columns (2) and (4) of Table V but their coefficients are not reported. Definitions of all variables are provided in the Table I. All regressions include year dummies. Robust standard errors clustered at the industry level are reported in parentheses. Industries are classified by Fama-French 48 industry groupings. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. Panel A : Interaction with Internal Monitoring Mechanisms Model 1 Model 2 Model 3 Model 4 F raud Detect|F raud F raud Detect|F raud F raud Detect|F raud F raud Detect|F raud (1) (2) (3) (4) (5) (6) (7) (8) A .1. Interaction with %_NonIndepDirectors FTA 1.081*** -0.818*** 0.957*** -0.707*** (0.352) (0.239) (0.276) (0.249) FTA*%_NonIndepDirectors -0.395 -0.356 (0.521) (0.475) AFTA 1.144*** -0.993*** 1.147*** -0.975*** (0.434) (0.325) (0.379) (0.325) AFTA*%_NonIndepDirectors -0.021 -0.078 (0.515) (0.480) %_NonIndepDirectors -1.810* 1.830** -1.962** 1.966** -1.777* 1.649 -1.795* 1.665* (0.926) (0.928) (0.880) (0.892) (1.020) (1.009) (0.988) (0.978) Observations 7871 7871 7871 7871 7871 7871 7871 7871 Wald Chi2 -932.03552 -932.00415 -931.14003 -931.11977 Prob> Chi2 0.00 0.00 0.00 0.00 A .2. Interaction with %_NonIndepDirectors_Audit FTA 0.884*** -0.762*** 0.957*** -0.707*** (0.290) (0.234) (0.276) (0.249) FTA*%_NonIndepDirectors_Audit 0.084 -0.356 (0.833) (0.475) AFTA 1.128*** -0.988*** 1.147*** -0.975*** (0.389) (0.326) (0.379) (0.325) AFTA*%_NonIndepDirectors_Audit 0.092 -0.078 (0.887) (0.480) %_NonIndepDirectors_Audit -2.100 1.109 -2.229 1.273 -1.960 1.033 -2.033 1.092 (2.872) (2.480) (2.924) (2.457) (2.935) (2.396) (2.875) (2.337) Observations 7871 7871 7871 7871 7871 7871 7871 7871 Wald Chi2 -932.51271 -932.00415 -931.13703 -931.11977 Prob> Chi2 0.00 0.00 0.00 0.00 Panel B : Interaction with External Monitoring Mechanisms FTA 0.872** -0.763*** 0.890*** -0.753** (0.441) (0.226) (0.276) (0.336) FTA*IOC 0.043 -0.032 (0.885) (0.699) AFTA 1.173** -0.989*** 1.136*** -0.933** (0.517) (0.319) (0.376) (0.388) AFTA*IOC -0.106 -0.152 (0.860) (0.712) IOC -1.559 1.511 -1.627 1.593 -1.670 1.580 -1.700 1.610 (1.611) (1.486) (1.591) (1.511) (1.540) (1.446) (1.544) (1.452) Observations 7871 7871 7871 7871 7871 7871 7871 7871 Wald Chi2 -932.51392 -932.51467 -931.12613 -931.1002 Prob> Chi2 0.00 0.00 0.00 0.00

#

($#

#

#

Appendix This Appendix contains the following regression estimates: Table A1: The regression used to construct our abnormal measures of CEO connectedness, AFTA

and AFTA_W. Table A2: First-stage IV regression estimates for the bivariate model estimation. Table A3: IV regression estimates for forced CEO turnover-fraud sensitivity. Table A4: IV regression estimates for fraud detection duration. Table A5: IV regression estimates for the number of executives charged. Table A6: Robustness test on whether results are driven by organizational changes. Table A7: Robustness test to clustering standard errors at the firm and CEO-firm pair level. Table A8: Re-estimation results using measures of CEO connectedness weighted by salaries and

bonuses for the bivariate regressions (Panel A); forced CEO turnover-fraud sensitivity (Panel B); fraud detection duration and the hazard ratio (Panel C); and the number of executives charged (Panel D).

Table A1. Regressions to Construct A F T A and A F T A_W . This table reports estimation results of the regression used to construct AFTA and AFTA_W. AFTA is the abnormal FTA, measured by the residual of the regression relating the fraction of top executives appointed (FTA) to factors affecting FTA. AFTA_W is the residual of the regression relating the value weighted fraction of top executives appointed (FTA_W) where the weight is the sum of salaries and bonuses each top four non-CEO executives earns in a given year. Definitions of all variables are provided in Table I. All regressions control for year fixed effects. Robust standard errors are in parentheses. Coefficients marked with *, **, and *** are significant at 10%, 5%, and 1%, respectively. F T A F T A_W (1) (2) CEO_Tenure 0.016*** 0.062*** (0.000) (0.001) Outside 0.005 0.037 (0.006) (0.025) Execsen -0.043*** -0.182*** (0.001) (0.005) FTA_1Y -0.076*** -0.317*** (0.006) (0.025) FTA_1Y_Unknown 0.397*** 1.681*** (0.063) (0.253) Unknown 0.328*** 1.273*** (0.021) (0.082) Constant 0.612*** 2.433*** (0.010) (0.039) Observations 21599 21599 Year FE Y Y Adjusted R2 0.16 0.16 #

(%#

#

#

Table A2. F irst-stage I V Regressions for the Bivariate Model Estimation.

This table reports the first-stage instrumental variable regression estimates for the IV regression in Table VI. The dependent variable is FTA in Columns (1) and (2); AFTA in Columns (3) and (4). The samples used in Columns (1)-(4) cover 1996 through 2006 and include all observations with necessary variables for each regression. The instrumental variables are Num_Death and Num_Retire. Columns (1) and (2) are estimated together as a system of simultaneous equations, as are Columns (3) and (4). Definitions of all variables are provided in Table I. All regressions control for year dummies. Robust standard errors are reported in parentheses. F-statistics (IVs) represents F statistics for an F-test that all the instrumental variables are jointly zero. Coefficients marked with *, **, and *** are significant at 10%, 5%, and 1%, respectively.

F T A A F T A (1) (2) (3) (4) Num_Death 0.055*** 0.055*** 0.054*** 0.054*** (0.018) (0.018) (0.016) (0.016) Num_Retire 0.028 0.028 0.016 0.016 (0.031) (0.031) (0.028) (0.029) Tobin's Q -0.001 -0.001 -0.000 -0.000 (0.002) (0.002) (0.002) (0.002) Ebitda/TA 0.070* 0.071* 0.066* 0.067* (0.041) (0.041) (0.038) (0.038) Leverage 0.037* 0.037* 0.047** 0.047** (0.021) (0.022) (0.020) (0.020) SalesGrowth_3Yr -0.000*** -0.000*** -0.000*** -0.000*** ## (0.000) (0.000) (0.000) (0.000) Log(TotalAssets) 0.006* 0.006* 0.008*** 0.008*** (0.003) (0.003) (0.003) (0.003) IndustryQ 0.002 ## 0.002 ## (0.005) ## (0.005) ##(IndustryQ)2 -0.000 ## -0.000 ## (0.001) ## (0.001) ##ICR 0.001 ## 0.002 ## (0.003) ## (0.003) ##CEO_OWN -0.012 ## -0.015 ## (0.016) ## (0.017) ##(CEO_OWN)2 0.015 ## 0.016 ## (0.044) ## (0.045) ##CEO_Founder -0.203*** -0.203*** -0.258*** -0.258*** (0.015) (0.015) (0.014) (0.014) CEO_Chair 0.024*** 0.024*** 0.018** 0.018** ## (0.009) (0.009) (0.008) (0.008) Ln(CEO_Age) -0.008 -0.008 -0.014 -0.014 (0.014) (0.014) (0.013) (0.013) CEO_Tenure 0.020*** 0.020*** 0.011*** 0.011*** (0.001) (0.001) (0.001) (0.001) Ln(BoardSize) 0.012 0.012 0.019 0.019 (0.017) (0.017) (0.016) (0.016) %_NonIndepDirectors -0.358*** -0.359*** -0.297*** -0.298*** (0.024) (0.024) (0.022) (0.022) Ln(BoardMeetings) 0.046*** 0.046*** 0.032*** 0.032*** (0.011) (0.011) (0.010) (0.010) %_NonIndepDirectors_Audit 0.364*** 0.364*** 0.323*** 0.323*** ## (0.060) (0.060) (0.055) (0.056) Ln(AuditComSize) 0.025 0.025 0.032** 0.032** (0.017) (0.017) (0.015) (0.016) StockVolatilities 1.142*** 1.147*** 0.302 0.308 (0.384) (0.387) (0.353) (0.356) IndustryLitigation 0.176 0.176 0.028 0.027 (0.175) (0.176) (0.161) (0.162) Ln(Analyst) 0.000 0.000 0.003 0.003 (0.005) (0.005) (0.005) (0.005) IOC 0.018 0.017 0.002 0.002 ## (0.034) (0.034) (0.031) (0.032) Constant 0.203*** 0.205*** -0.001 0.001 ## (0.071) (0.071) (0.065) (0.065) Year Dummies Y Y Y Y Observations 7871 7871 7871 7871 Chi2 1421.09 1400.63 826.78 810.48 F-statistics (IVs) 10.34 10.20 11.61 **,(*#Prob > F (IVs) 0.0057 0.0061 0.0030 ),))!!#

(&#

#

Table A3. I V Regressions for Forced C E O Turnover-F raud Sensitivity. Columns (1)-(2) and (4)-(5) report the first-stage IV regression estimates, where the dependent variable is FTA in Column (1), Fraud_t-2-t*FTA in Column (2), AFTA in Column (4), and Fraud_t-2-t *AFTA in Column (5). The instrumental variables are Num_Death and Num_Retire. Columns (3) and (6) report the second stage regression results. F-statistics (IVs) represent F statistics for an F-test that all the instrumental variables are jointly zero. All regressions control for firm- and year fixed effects. The sample covers the period 1996 through 2006. Definitions of all variables are provided in Table I. Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. 1st Stage 2nd Stage 1st Stage 2nd Stage F T A F raud_t-2-t*F T A Forced F T A F raud_t-2-t*F T A Forced (1) (2) (3) (4) (5) (6) Fraud_t-2-t *FTA_Hat -0.689*** (0.197) Fraud_t-2-t *AFTA_Hat -0.303 (0.642) FTA_Hat -0.255 (0.175) AFTA_Hat 0.549 (0.217) Fraud_t-2-t 0.025 0.495*** 0.402*** 0.020 0.070*** 0.020 (0.023) (0.026) (0.107) (0.021) (0.025) (0.047) Tobin's Q -0.004** 0.000 -0.002 -0.003* 0.001 -0.002 (0.002) (0.001) (0.002) (0.002) (0.001) (0.002) Ebitda/TA -0.053 -0.025 -0.058 -0.047 -0.018 -0.032 (0.086) (0.026) (0.050) (0.078) (0.022) (0.053) Leverage -0.045 -0.021 -0.040 -0.054 -0.015 -0.022 (0.056) (0.014) (0.036) (0.051) (0.013) (0.037) SalesGrowth_3Yr 0.000 0.000 0.000 -0.000 0.000 -0.000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Log(TotalAssets) 0.011 -0.001 0.009 0.015 -0.000 0.011 (0.019) (0.004) (0.011) (0.017) (0.004) (0.012) IndustryQ -0.107 -0.032 -0.063 -0.083 -0.020 -0.027 (0.071) (0.022) (0.051) (0.062) (0.020) (0.053) (IndustryQ)2 0.014 0.005 0.011 0.009 0.003 0.006 (0.016) (0.005) (0.011) (0.014) (0.004) (0.011) ICR 0.136 0.075** 0.029 0.144 0.061** -0.048 (0.154) (0.032) (0.079) (0.143) (0.027) (0.093) CEO_OWN -0.577 -0.025 -0.078 -0.736** -0.016 -0.128 (0.368) (0.058) (0.222) (0.361) (0.050) (0.256) (CEO_OWN)2 0.554 0.032 -0.139 0.853 0.027 -0.059 (0.804) (0.092) (0.323) (0.754) (0.075) (0.361) CEO_Founder -0.360*** -0.010 0.030 -0.406*** -0.008 0.009 (0.056) (0.014) (0.070) (0.055) (0.011) (0.095) CEO_Chair 0.029** -0.002 0.001 0.024* -0.001 0.003 (0.014) (0.003) (0.011) (0.013) (0.003) (0.011) Ln(CEO_Age) 0.063 0.006 0.101*** 0.067 0.006 0.098*** (0.043) (0.009) (0.031) (0.043) (0.008) (0.034) CEO_Tenure 0.036*** 0.002*** 0.000 0.025*** 0.001** -0.003 (0.003) (0.001) (0.006) (0.003) (0.000) (0.006) Ln(BoardSize) -0.049 -0.020* -0.044** -0.019 -0.012 -0.016 (0.035) (0.011) (0.021) (0.031) (0.009) (0.022) %_NonIndepDirectors -0.153*** -0.000 -0.043 -0.133*** 0.004 -0.046 (0.046) (0.013) (0.041) (0.042) (0.012) (0.042) Ln(BoardMeetings) -0.008 -0.012* 0.036*** -0.008 -0.011** 0.050*** (0.017) (0.006) (0.012) (0.016) (0.005) (0.016) %_NonIndepDirectors_Audit 0.108 -0.004 -0.015 0.095 -0.006 -0.007 (0.073) (0.018) (0.054) (0.064) (0.016) (0.054) Ln(AuditComSize) 0.014 0.005 0.006 0.004 0.005 -0.003 (0.019) (0.005) (0.013) (0.017) (0.005) (0.013) Ln(Analyst) -0.000 -0.000 -0.018* 0.008 0.003 -0.017 (0.015) (0.004) (0.011) (0.014) (0.003) (0.011) IOC -0.038 0.031** 0.082** -0.047 0.025* 0.041 (0.060) (0.015) (0.036) (0.054) (0.013) (0.037) StockVolatilities -0.492 -0.185 0.212 -0.569 -0.157 0.387 (0.595) (0.206) (0.451) (0.532) (0.175) (0.471) IndustryLitigation -0.062 0.009 0.298* -0.059 0.011 0.285* (0.166) (0.062) (0.167) (0.146) (0.051) (0.167) Jail -0.024 0.009 0.132 -0.054 0.018 0.106 (0.285) (0.278) (0.137) (0.257) (0.196) (0.138) Num_Death 0.065** -0.004 0.060* -0.003 (0.030) (0.003) (0.032) (0.002) Num_Retire 0.132 0.001 0.117 -0.000 (0.087) (0.005) (0.085) (0.003) Fraudt-2*Num_Death 0.244*** 0.254*** 0.099*** 0.146*** (0.039) (0.028) (0.037) (0.026) Fraudt-2*Num_Retire -0.111 0.193*** -0.174** -0.171*** (0.084) (0.019) (0.081) (0.017) Constant 0.198 0.052 -0.149 -0.159 0.017 -0.293** (0.187) (0.048) (0.121) (0.175) (0.042) (0.129) Firm FE & Year FE Y Y Y Y Y Y Observations 7871 7871 7871 7871 7871 7871 Adj-R2 0.61 0.77 0.10 0.61 0.30 0.10 F-statistics (IVs) 26.1 30.12 12.13 607.63 Prob > F (IVs) 0.0000 0.0000 0.0000 0.0000

#

('#

#

#

Table A4. I V Regressions for F raud Detection Duration. Columns (1) and (3) report the first-stage regression estimates, where the dependent variable is FTA and AFTA, respectively. The instrumental variables are Num_Death and Num_Retire. Columns (2) and (4) report the second stage regression results for fraud detection duration. F-statistics (IVs) represent F statistics for an F-test that all the instrumental variables are jointly zero. The sample covers 246 detected fraud cases for the period 1996 through 2006. CEO connectedness and all control variables are their average values over the fraud period. Definitions of all variables are provided in Table I. All regressions control for industry dummies. Robust standard errors clustered at the industry level are reported in parentheses. Industries are classified by Fama-French 48 industry groupings. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. 1st Stage 2nd Stage 1st Stage 2nd Stage F T A Ln(Duration+1) A F T A Ln(Duration+1) (1) (2) (3) (4) FTA_Hat 1.150*** [0.116] AFTA_Hat 1.602*** [0.232] Log(TotalAssets) 0.048*** -0.100** 0.050*** -0.125** [0.017] [0.047] [0.017] [0.050] Tobin's Q 0.010 0.013 0.007 0.014 [0.009] [0.014] [0.009] [0.014] Leverage -0.187 1.002*** -0.118 0.978*** [0.152] [0.348] [0.169] [0.354] SalesGrowth_3Yr -0.000 0.001*** -0.000 0.001*** [0.000] [0.000] [0.000] [0.000] StockVolatilities -1.135 -16.555*** -0.879 -16.445*** [1.614] [5.295] [1.638] [5.244] IndustryLitigation 1.189 0.369 0.191 1.427 [1.548] [3.868] [1.333] [3.874] Ln(Analyst) 0.028 -0.099 0.025 -0.108 [0.024] [0.079] [0.023] [0.079] Ln(CEO_Age) 0.110* -0.350** 0.025 -0.264 [0.059] [0.171] [0.071] [0.172] Num_Death 0.323*** 0.264*** [0.010] [0.010] Num_Retire -0.004 0.008 [0.093] [0.058] Constant 0.429** 7.001*** 0.217 7.150*** [0.211] [0.711] [0.249] [0.700] Industry Dummies Y Y Y Y Observations 246 246 246 246 Adj-R2 0.09 0.08 0.09 0.08 F-statistics (IVs) 436.02 518.19 Prob > F (IVs) 0.0000 0.0000 #

")#

#

#

Table A5. I V Regressions for the Number of Executives Charged. This table reports the IV regression estimates for the number of executives charged. Columns (1) and (3) report the first-stage estimates, where the dependent variable is FTA and AFTA, respectively. The instrumental variable is Num_Retire. Columns (2) and (4) report the second stage estimation results. F-statistics (IVs) represent F statistics for an F-test that all the instrumental variables are jointly zero. The sample covers 275 detected fraud cases over the period 1996 through 2006. CEO connectedness and all control variables are their average values over the fraud period. Definitions of all variables are provided in Table I. All regressions control for industry dummies. Robust standard errors clustered at the industry level are reported in parentheses. Industries are classified by Fama-French 48 industry groupings. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. 1st Stage 2nd Stage 1st Stage 2nd Stage F T A Ln(Num_Charged+1) A F T A Ln(Num_Charged+1) (1) (2) (3) (4) FTA_Hat 1.500*** (0.430) AFTA_Hat 4.439*** (1.273) Tobin's Q 0.003 0.001 0.002 -0.003 (0.006) (0.002) (0.006) (0.003) SalesGrowth_3Yr -0.001*** 0.000* -0.000** 0.002*** (0.000) (0.000) (0.000) (0.001) Leverage -0.228 0.539*** -0.128 0.763*** (0.169) (0.170) (0.171) (0.228) Accounting -0.024 0.092** -0.003 0.067* (0.086) (0.034) (0.069) (0.035) Operating 0.030 0.025 0.050 -0.150** (0.064) (0.038) (0.055) (0.072) Executive 0.063 -0.075 0.072 -0.302*** (0.059) (0.050) (0.055) (0.105) Num_Retire 0.060 0.020 (0.080) (0.059) Constant 0.545*** 0.117 0.160 0.222 (0.177) (0.245) (0.157) (0.216) Industry Dummies Y Y Y Y Observations 275 275 275 275 Adj-R2 -0.04 0.11 -0.03 0.11 F-statistics (IVs) 0.56 0.12 Prob > F (IVs) 0.4573 0.7350

#

"*#

#

#Table A6. Robustness to A lternative Story: A re Results Driven by O rganizational Changes? This table tests whether the relation between CEO connectedness and corporate frauds estimated by the bivariate model is driven by organizational changes, as proxied by MA, DS, and Segment. MA and DS are the number of mergers and acquisitions (MA) and divestitures and spinoffs (DS) completed by a firm in the previous year. Segment is the number of segments a firm has in a given year. Fraud and detection regression specifications are identical to those in Table V. Coefficients marked with *, ** and *** are significant at 10%, 5%, 1%, respectively. Model 1 Model 2 F raud Detect|F raud F raud Detect|F raud (1) (2) (3) (4) FTA 1.061*** -1.114*** (0.329) (0.337) AFTA 1.203*** -1.291*** (0.388) (0.360) Segments -0.011* -0.012 (0.007) (0.008) MA 0.038 0.040 (0.025) (0.026) DS -0.009 -0.010 (0.026) (0.027) Tobin's Q -0.376*** 0.355** -0.385*** 0.364** (0.097) (0.142) (0.107) (0.162) Ebitda/TA -2.201 1.678 -1.919 1.399 (1.647) (1.564) (1.503) (1.431) Leverage -1.364 1.607 -1.279 1.542 (1.851) (1.906) (1.677) (1.762) SalesGrowth_3Yr 0.016*** -0.009* 0.016*** -0.010 (0.004) (0.005) (0.004) (0.006) Log(TotalAssets) -0.206 0.295** -0.192 0.287** (0.138) (0.119) (0.139) (0.122) IndustryQ 0.139 0.155 (0.425) (0.440) (IndustryQ)2 0.017 0.016 (0.091) (0.092) ICR 0.186 0.200 (0.192) (0.209) CEO_OWN -0.462 -0.538 (1.236) (1.309) (CEO_OWN)2 -0.038 0.214 (3.728) (4.058) CEO_Founder 0.166 -0.215 0.254 -0.317 (0.785) (0.822) (0.606) (0.650) CEO_Chair -0.097 0.069 -0.121 0.092 (0.387) (0.411) (0.348) (0.370) Ln(CEO_Age) 0.106 -0.351 0.078 -0.330 (0.721) (0.729) (0.694) (0.704) CEO_Tenure -0.003 0.015 0.010 0.002 (0.022) (0.026) (0.020) (0.022) Ln(BoardSize) -1.759 2.094* -1.861 2.223** (1.251) (1.099) (1.323) (1.093) %_NonIndepDirectors -0.181 0.357 -0.267 0.453 (1.058) (1.257) (1.116) (1.309) Ln(BoardMeetings) 0.308 -0.164 0.306 -0.158 (0.484) (0.521) (0.496) (0.544) %_NonIndepDirectors_Audit -2.474 1.765 -2.365 1.653 (2.610) (2.635) (2.566) (2.531) Ln(AuditComSize) 1.685** -1.432** 1.518** -1.265* (0.721) (0.674) (0.706) (0.651) Ln(Analyst) 0.277 -0.184 0.255 -0.156 (0.235) (0.239) (0.218) (0.219) IOC -1.024 0.884 -1.067 0.923 (1.427) (1.464) (1.323) (1.343) StockVolatilities -17.472 33.326*** -15.781 31.954*** (16.485) (10.676) (16.578) (10.325) IndustryLitigation 15.963*** -11.521*** 15.692*** -11.292*** (4.597) (3.553) (4.674) (3.855) Constant 1.892 -3.256 2.439 -3.929 (3.042) (2.291) (3.392) (2.553) Year Dummies Y Y Y Y Observations 6948 6948 6948 6948 Wald Chi2 -854.98447 -853.51582 Prob> Chi2 0.00 0.00

#

"+#

#

#Table A7. Robustness to C lustering Standard E r rors at the F irm- and C E O-firm Pair Level.

This table re-estimates the relation between CEO connectedness and corporate frauds while clustering standard errors at different levels. Panels A and B estimate the relation while clustering at the firm- and CEO-firm pair level. Fraud and detection regression specifications are identical to those in Table V. All regressions include year dummies effects. Coefficients marked with *, ** and *** are significant at 10%, 5%, 1%, respectively. Panel A : C luster ing at the F irm L evel Panel B : C luster ing at the C E O-F irm Pair Level Model 1 Model 2 Model 3 Model 4 F raud Detect|F raud F raud Detect|F raud F raud Detect|F raud F raud Detect|F raud (1) (2) (3) (4) (5) (6) (7) (8) FTA 1.023** -0.737* 1.023** -0.737* (0.425) (0.407) (0.416) (0.398) AFTA 1.273*** -0.949** 1.273*** -0.949** (0.422) (0.455) (0.415) (0.444) Tobin's Q -0.282*** 0.193* -0.297*** 0.206* -0.282*** 0.193* -0.297*** 0.206** (0.089) (0.108) (0.088) (0.105) (0.088) (0.107) (0.087) (0.104) Ebitda/TA -2.377 1.238 -2.150 1.094 -2.377 1.238 -2.150 1.094 (1.715) (1.045) (1.774) (1.079) (1.628) (1.015) (1.691) (1.050) Leverage -1.421 1.354 -1.504 1.420 -1.421 1.354 -1.504 1.420 (1.708) (1.440) (1.419) (1.188) (1.646) (1.383) (1.376) (1.148) SalesGrowth_3Yr 0.014** -0.003 0.014** -0.004 0.014** -0.003 0.014** -0.004 (0.006) (0.004) (0.006) (0.004) (0.006) (0.004) (0.006) (0.004) Log(TotalAssets) -0.063 0.176 -0.056 0.171* -0.063 0.176 -0.056 0.171* (0.185) (0.112) (0.173) (0.101) (0.179) (0.111) (0.168) (0.100) IndustryQ 0.089 0.102 0.089 0.102 (0.580) (0.543) (0.568) (0.534) (IndustryQ)2 0.046 0.043 0.046 0.043 (0.168) (0.155) (0.163) (0.151) ICR 0.351 0.331 0.351 0.331 (0.583) (0.545) (0.564) (0.527) CEO_OWN -0.554 -0.590 -0.554 -0.590 (2.330) (2.285) (2.312) (2.266) (CEO_OWN)2 -1.390 -0.993 -1.390 -0.993 (7.240) (6.932) (7.229) (6.919) CEO_Founder 0.291 -0.220 0.335 -0.263 0.291 -0.220 0.335 -0.263 (1.154) (0.807) (0.969) (0.684) (1.117) (0.776) (0.943) (0.660) CEO_Chair -0.441 0.299 -0.426 0.291 -0.441 0.299 -0.426 0.291 (0.500) (0.317) (0.466) (0.301) (0.483) (0.304) (0.452) (0.291) Ln(CEO_Age) 0.859 -0.880 0.852 -0.874 0.859 -0.880 0.852 -0.874 (0.787) (0.594) (0.725) (0.559) (0.774) (0.578) (0.716) (0.546) CEO_Tenure -0.014 0.019 -0.002 0.011 -0.014 0.019 -0.002 0.011 (0.043) (0.034) (0.034) (0.030) (0.041) (0.033) (0.033) (0.029) Ln(BoardSize) -2.434* 1.940** -2.530** 2.017** -2.434* 1.940** -2.530** 2.017** (1.308) (0.939) (1.129) (0.886) (1.271) (0.922) (1.102) (0.872) %_NonIndepDirectors -0.878 0.958 -0.902 0.967 -0.878 0.958 -0.902 0.967 (1.156) (1.034) (1.120) (0.999) (1.129) (1.006) (1.096) (0.974) Ln(BoardMeetings) 0.048 0.107 0.068 0.091 0.048 0.107 0.068 0.091 (0.650) (0.526) (0.549) (0.454) (0.636) (0.516) (0.541) (0.448) %_NonIndepDirectors_Audit -1.440 0.392 -1.400 0.399 -1.440 0.392 -1.400 0.399 (2.364) (1.912) (2.232) (1.730) (2.290) (1.811) (2.167) (1.647) Ln(AuditComSize) 0.602 -0.271 0.509 -0.203 0.602 -0.271 0.509 -0.203 (0.563) (0.405) (0.541) (0.384) (0.562) (0.402) (0.540) (0.382) Ln(Analyst) -0.172 0.192 -0.210 0.218 -0.172 0.192 -0.210 0.218 (0.324) (0.214) (0.289) (0.185) (0.318) (0.209) (0.283) (0.180) IOC -1.676 1.203 -1.746 1.258 -1.676 1.203 -1.746 1.258 (1.961) (1.821) (1.658) (1.553) (1.914) (1.774) (1.628) (1.522) StockVolatilities -20.536* 29.987*** -18.835 28.548*** -20.536* 29.987*** -18.835 28.548*** (11.685) (8.829) (11.847) (8.753) (11.745) (8.772) (11.894) (8.690) IndustryLitigation 24.253 -9.335** 22.876 -8.846** 24.253 -9.335** 22.876 -8.846** (15.830) (4.359) (15.684) (3.821) (15.214) (4.192) (15.044) (3.702) Constant 4.716 -5.028** 5.293* -5.432*** 4.716 -5.028** 5.293* -5.432*** (3.121) (1.998) (2.874) (1.857) (3.065) (1.988) (2.837) (1.856) Year Dummies Y Y Y Y Y Y Y Y Observations 7871 7871 7871 7871 7871 7871 7871 7871 Wald Chi2 -1033.7673 -1031.9532 -1033.7673 -1031.9532 Prob> Chi2 0.00 0.00 0.00 0.00

#

"!#

#

#

#Table A8. Re-estimation Results Using M easures of C E O Connectedness W eighted by Salaries and Bonuses.

Panel A : Bivariate Regressions This panel re-estimates the bivariate model using measures of CEO connectedness weighted by non-CEO weight to higher-ranked executives. FTA_W in Columns (1) and (2) is FTA weighted by non-CEO Columns (3) and (4) is the residual based on FTA_W. Columns (1) and (3) report estimation results for the fraud incidence. Columns (2) and (4) report estimation results for fraud detection. Control variables are identical to those in Table V. All regressions include year dummies. Robust standard errors clustered at the industry level are reported in parentheses. Industries are classified by Fama-French 48 industry groupings. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. Model 1 Model 2 F raud Detect|F raud F raud Detect|F raud (1) (2) (3) (4) FTA_W 1.123*** -0.768** (0.363) (0.324) AFTA_W 1.389*** -0.971*** (0.499) (0.361) Tobin's Q -0.292*** 0.204** -0.309*** 0.216** (0.086) (0.093) (0.093) (0.095) Ebitda/TA -2.007 1.034 -1.778 0.877 (1.482) (0.930) (1.472) (0.910) Leverage -1.623 1.492 -1.719 1.541 (1.988) (1.611) (2.012) (1.661) SalesGrowth_3Yr 0.013 -0.003 0.013 -0.003 (0.009) (0.003) (0.011) (0.003) Log(TotalAssets) -0.104 0.197* -0.103 0.191 (0.237) (0.117) (0.272) (0.122) IndustryQ 0.191 0.207 (0.504) (0.481) (IndustryQ)2 0.017 0.013 (0.146) (0.142) ICR 0.261 0.231 (0.489) (0.508) CEO_OWN -0.277 -0.326 (1.363) (1.268) (CEO_OWN)2 -1.854 -1.313 (5.657) (5.149) CEO_Founder 0.147 -0.119 0.190 -0.157 (1.358) (0.937) (1.421) (0.950) CEO_Chair -0.211 0.148 -0.190 0.132 (0.492) (0.346) (0.505) (0.340) Ln(CEO_Age) 0.684 -0.743 0.657 -0.708 (0.720) (0.639) (0.712) (0.639) CEO_Tenure -0.009 0.014 0.002 0.006 (0.032) (0.032) (0.028) (0.033) Ln(BoardSize) -2.143 1.764 -2.229 1.808 (1.563) (1.101) (1.696) (1.115) %_NonIndepDirectors -0.646 0.787 -0.660 0.773 (1.096) (1.021) (1.088) (0.984) Ln(BoardMeetings) 0.171 0.004 0.174 -0.001 (0.502) (0.464) (0.453) (0.437) %_NonIndepDirectors_Audit -1.446 0.464 -1.320 0.413 (2.679) (2.028) (2.689) (1.889) Ln(AuditComSize) 0.555 -0.253 0.438 -0.168 (0.758) (0.523) (0.742) (0.468) StockVolatilities -17.799 27.435*** -16.323 25.696*** (11.073) (8.885) (14.073) (8.756) IndustryLitigation 21.313 -8.513*** 19.399 -7.611** (18.451) (3.137) (21.587) (3.197) Ln(Analyst) -0.228 0.227 -0.276 0.254 (0.335) (0.205) (0.293) (0.178) IOC -1.827 1.329 -1.901 1.366 (1.525) (1.478) (1.335) (1.370) Constant 3.873 -4.341** 4.657* -4.807*** (2.538) (1.901) (2.629) (1.768) Year Dummies Y Y Y Y Observations 7871 7871 7871 7871 Wald Chi2 -1029.1204 -1027.095 Prob> Chi2 0.00 0.00

#

"(#

#

#

Panel B . Forced C E O Turnover-F raud Sensitivity. This panel re-estimates the forced CEO turnover-fraud sensitivity using measures of CEO connectedness weighted by non-CEO execubonuses to give greater weight to higher-ranked executives. FTA_W in Columns (1) and (2) is FTA weighted by non-CEO bonuses; AFTA_W in Columns (3) and (4) is the residual based on FTA_W. Columns (1) and (2) report results by the OLS estimation. Columns (3) and (4) report results estimated by conditional logit regression. Control variables are identical to those in Table VII. OLS regressions control for firm- and year fixed effects and logit regressions control for year dummies. Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. Forced_C E O_Turnover O LS C logit (1) (2) (3) (4) Fraud_t-2-t*FTA_W -0.221*** -3.049** (0.063) (1.524) Fraud_t-2-t*AFTA_W -0.269*** -4.339** (0.072) (1.839) FTA_W -0.100*** -2.671*** (0.015) (0.542) AFTA_W -0.121*** -2.777*** (0.018) (0.592) Fraud_t-2-t 0.163*** 0.076*** 2.302*** 1.281** (0.044) (0.023) (0.817) (0.623) Tobin's Q -0.001 -0.001 0.020 0.020 (0.002) (0.002) (0.038) (0.037) Ebitda/TA -0.033 -0.032 -0.381 -0.344 (0.050) (0.049) (1.877) (1.839) Leverage -0.019 -0.021 -0.036 0.042 (0.035) (0.035) (1.427) (1.447) SalesGrowth_3Yr 0.000 0.000 0.001 0.001 (0.000) (0.000) (0.002) (0.001) Log(TotalAssets) 0.006 0.007 -0.238 -0.238 (0.011) (0.011) (0.523) (0.519) IndustryQ -0.031 -0.029 -1.566 -1.581 (0.046) (0.046) (2.519) (2.523) (IndustryQ)2 0.007 0.006 0.384 0.392 (0.010) (0.010) (0.554) (0.554) ICR -0.017 -0.014 -6.769* -6.843 (0.077) (0.076) (4.107) (4.214) CEO_OWN 0.034 0.005 19.105 19.115 (0.198) (0.201) (13.837) (13.754) (CEO_OWN)2 -0.262 -0.216 -100.968** -105.862** (0.302) (0.304) (44.485) (46.027) CEO_Founder 0.094*** 0.080*** 5.151*** 5.129*** (0.024) (0.024) (1.787) (1.816) CEO_Chair -0.002 -0.002 -0.221 -0.220 (0.010) (0.010) (0.312) (0.314) Ln(CEO_Age) 0.087*** 0.089*** 1.674** 1.709** (0.027) (0.027) (0.787) (0.816) CEO_Tenure -0.006*** -0.007*** -0.478*** -0.506*** (0.001) (0.001) (0.138) (0.136) Ln(BoardSize) -0.025 -0.021 -1.463 -1.215 (0.020) (0.020) (0.920) (0.908) %_NonIndepDirectors -0.015 -0.014 0.547 0.690 (0.033) (0.033) (1.401) (1.390) Ln(BoardMeetings) 0.042*** 0.041*** 1.130** 1.064** (0.012) (0.012) (0.465) (0.463) %_NonIndepDirectors_Audit -0.031 -0.031 -2.444 -2.510 (0.050) (0.050) (2.103) (2.122) Ln(AuditComSize) 0.000 -0.000 0.387 0.434 (0.013) (0.013) (0.557) (0.552) Ln(Analyst) -0.018* -0.016 -0.949** -0.923** (0.010) (0.010) (0.452) (0.453) IOC 0.076** 0.074** 3.510** 3.368** (0.034) (0.034) (1.548) (1.555) StockVolatilities 0.445 0.429 24.132 24.720 (0.432) (0.431) (17.856) (17.801) IndustryLitigation 0.287* 0.288* 7.353 7.115 (0.165) (0.165) (4.508) (4.529) Jail 0.136** 0.135** 5.033*** 4.872** (0.068) (0.063) (1.863) (1.948) Constant -0.210* -0.254** (0.117) (0.118) Firm FE & Year FE Y Y N N Year Dummies N N Y Y Observations 7871 7871 1195 1195 Adj-R2 (Pseudo R2) 0.12 0.13 (0.45) (0.45)

""#

#

#

Panel C . F raud Detection Duration and the Hazard Ratio. This panel re-estimates the relation for fraud detection duration and the hazard ratio using measures of CEO connectedness weighted by non-CEO o give greater weight to higher-ranked executives. FTA_W in Columns (1) and (3) is FTA weighted by non-CEO in Columns (2) and (4) is the residual based on FTA_W. Columns (1) and (2) report the results for the OLS estimation for fraud detection duration. Columns (3) and (4) report the results for the Cox regressions. Control variables are identical to those in Table VIII. All regressions include industry dummies. Robust standard errors clustered at the industry level are reported in parentheses. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. Ln(Duration+1) _t (1) (2) (3) (4) FTA_W 0.392** -0.784*** (0.180) (0.238) AFTA_W 0.211 -0.444** (0.155) (0.198) Log(TotalAssets) -0.063 -0.055 0.031 0.019 (0.047) (0.048) (0.059) (0.060) Tobin's Q 0.020 0.023* -0.035 -0.041 (0.013) (0.013) (0.043) (0.047) Leverage 0.864** 0.813** -1.136** -1.058** (0.340) (0.342) (0.555) (0.536) SalesGrowth_3Yr 0.001** 0.001** -0.001 -0.001 (0.000) (0.000) (0.001) (0.001) StockVolatilities -17.399*** -17.673*** 19.187** 19.020** (5.441) (5.413) (8.686) (8.920) IndustryLitigation 1.269 1.681 -8.828 -9.575 (4.055) (3.941) (6.138) (6.168) Ln(Analyst) -0.077 -0.071 0.170** 0.155* (0.077) (0.076) (0.084) (0.082) Ln(CEO_Age) -0.260 -0.226 0.291 0.198 (0.182) (0.177) (0.325) (0.323) Constant 7.161*** 7.244*** (0.574) (0.598) Industry Dummies Y Y Y Y Observations 246 246 246 246 Adj-R2 0.10 0.09 Wald Chi2 28252.17 108933.07

#

"$#

#

#

Panel D . Number of Executives Charged. This panel re-estimates the relation for the number of executives charged using measures of CEO connectedness weighted by non-CEO -ranked executives. FTA_W in Columns (1) is FTA weighted by non-CEO bonuses; AFTA_W in Columns (2) is the residual based on FTA_W. Control variables are identical to those in Table XI. All regressions include industry dummies. Robust standard errors clustered at the industry level are reported in parentheses. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%, respectively. Ln(Num_Charged+1) (1) (2) FTA_W 0.121*** (0.034) AFTA_W 0.112** (0.042) Tobin's Q 0.004** 0.004** (0.002) (0.002) SalesGrowth_3Yr -0.000*** -0.000*** (0.000) (0.000) Leverage 0.218** 0.204** (0.084) (0.086) Accounting 0.058 0.055 (0.035) (0.035) Operating 0.067** 0.065* (0.033) (0.034) Executive 0.013 0.013 (0.038) (0.039) Constant 0.863*** 0.910*** (0.071) (0.065) Industry Dummies Y Y Observations 275 275 Adj-R2 0.13 0.13 #